1 /* SPDX-License-Identifier: GPL-2.0 */ 2 /* 3 * fs/f2fs/f2fs.h 4 * 5 * Copyright (c) 2012 Samsung Electronics Co., Ltd. 6 * http://www.samsung.com/ 7 */ 8 #ifndef _LINUX_F2FS_H 9 #define _LINUX_F2FS_H 10 11 #include <linux/uio.h> 12 #include <linux/types.h> 13 #include <linux/page-flags.h> 14 #include <linux/buffer_head.h> 15 #include <linux/slab.h> 16 #include <linux/crc32.h> 17 #include <linux/magic.h> 18 #include <linux/kobject.h> 19 #include <linux/sched.h> 20 #include <linux/cred.h> 21 #include <linux/vmalloc.h> 22 #include <linux/bio.h> 23 #include <linux/blkdev.h> 24 #include <linux/quotaops.h> 25 #include <linux/part_stat.h> 26 #include <crypto/hash.h> 27 28 #include <linux/fscrypt.h> 29 #include <linux/fsverity.h> 30 31 #ifdef CONFIG_F2FS_CHECK_FS 32 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 33 #else 34 #define f2fs_bug_on(sbi, condition) \ 35 do { \ 36 if (WARN_ON(condition)) \ 37 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 38 } while (0) 39 #endif 40 41 enum { 42 FAULT_KMALLOC, 43 FAULT_KVMALLOC, 44 FAULT_PAGE_ALLOC, 45 FAULT_PAGE_GET, 46 FAULT_ALLOC_NID, 47 FAULT_ORPHAN, 48 FAULT_BLOCK, 49 FAULT_DIR_DEPTH, 50 FAULT_EVICT_INODE, 51 FAULT_TRUNCATE, 52 FAULT_READ_IO, 53 FAULT_CHECKPOINT, 54 FAULT_DISCARD, 55 FAULT_WRITE_IO, 56 FAULT_MAX, 57 }; 58 59 #ifdef CONFIG_F2FS_FAULT_INJECTION 60 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 61 62 struct f2fs_fault_info { 63 atomic_t inject_ops; 64 unsigned int inject_rate; 65 unsigned int inject_type; 66 }; 67 68 extern const char *f2fs_fault_name[FAULT_MAX]; 69 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 70 #endif 71 72 /* 73 * For mount options 74 */ 75 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 76 #define F2FS_MOUNT_DISCARD 0x00000004 77 #define F2FS_MOUNT_NOHEAP 0x00000008 78 #define F2FS_MOUNT_XATTR_USER 0x00000010 79 #define F2FS_MOUNT_POSIX_ACL 0x00000020 80 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 81 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 82 #define F2FS_MOUNT_INLINE_DATA 0x00000100 83 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 84 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 85 #define F2FS_MOUNT_NOBARRIER 0x00000800 86 #define F2FS_MOUNT_FASTBOOT 0x00001000 87 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 88 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 89 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 90 #define F2FS_MOUNT_USRQUOTA 0x00080000 91 #define F2FS_MOUNT_GRPQUOTA 0x00100000 92 #define F2FS_MOUNT_PRJQUOTA 0x00200000 93 #define F2FS_MOUNT_QUOTA 0x00400000 94 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 95 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 96 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 97 #define F2FS_MOUNT_NORECOVERY 0x04000000 98 #define F2FS_MOUNT_ATGC 0x08000000 99 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000 100 #define F2FS_MOUNT_GC_MERGE 0x20000000 101 102 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 103 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 104 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 105 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 106 107 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 108 typecheck(unsigned long long, b) && \ 109 ((long long)((a) - (b)) > 0)) 110 111 typedef u32 block_t; /* 112 * should not change u32, since it is the on-disk block 113 * address format, __le32. 114 */ 115 typedef u32 nid_t; 116 117 #define COMPRESS_EXT_NUM 16 118 119 struct f2fs_mount_info { 120 unsigned int opt; 121 int write_io_size_bits; /* Write IO size bits */ 122 block_t root_reserved_blocks; /* root reserved blocks */ 123 kuid_t s_resuid; /* reserved blocks for uid */ 124 kgid_t s_resgid; /* reserved blocks for gid */ 125 int active_logs; /* # of active logs */ 126 int inline_xattr_size; /* inline xattr size */ 127 #ifdef CONFIG_F2FS_FAULT_INJECTION 128 struct f2fs_fault_info fault_info; /* For fault injection */ 129 #endif 130 #ifdef CONFIG_QUOTA 131 /* Names of quota files with journalled quota */ 132 char *s_qf_names[MAXQUOTAS]; 133 int s_jquota_fmt; /* Format of quota to use */ 134 #endif 135 /* For which write hints are passed down to block layer */ 136 int whint_mode; 137 int alloc_mode; /* segment allocation policy */ 138 int fsync_mode; /* fsync policy */ 139 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 140 int bggc_mode; /* bggc mode: off, on or sync */ 141 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 142 block_t unusable_cap_perc; /* percentage for cap */ 143 block_t unusable_cap; /* Amount of space allowed to be 144 * unusable when disabling checkpoint 145 */ 146 147 /* For compression */ 148 unsigned char compress_algorithm; /* algorithm type */ 149 unsigned char compress_log_size; /* cluster log size */ 150 unsigned char compress_level; /* compress level */ 151 bool compress_chksum; /* compressed data chksum */ 152 unsigned char compress_ext_cnt; /* extension count */ 153 int compress_mode; /* compression mode */ 154 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 155 }; 156 157 #define F2FS_FEATURE_ENCRYPT 0x0001 158 #define F2FS_FEATURE_BLKZONED 0x0002 159 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 160 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 161 #define F2FS_FEATURE_PRJQUOTA 0x0010 162 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 163 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 164 #define F2FS_FEATURE_QUOTA_INO 0x0080 165 #define F2FS_FEATURE_INODE_CRTIME 0x0100 166 #define F2FS_FEATURE_LOST_FOUND 0x0200 167 #define F2FS_FEATURE_VERITY 0x0400 168 #define F2FS_FEATURE_SB_CHKSUM 0x0800 169 #define F2FS_FEATURE_CASEFOLD 0x1000 170 #define F2FS_FEATURE_COMPRESSION 0x2000 171 172 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 173 ((raw_super->feature & cpu_to_le32(mask)) != 0) 174 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 175 #define F2FS_SET_FEATURE(sbi, mask) \ 176 (sbi->raw_super->feature |= cpu_to_le32(mask)) 177 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 178 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 179 180 /* 181 * Default values for user and/or group using reserved blocks 182 */ 183 #define F2FS_DEF_RESUID 0 184 #define F2FS_DEF_RESGID 0 185 186 /* 187 * For checkpoint manager 188 */ 189 enum { 190 NAT_BITMAP, 191 SIT_BITMAP 192 }; 193 194 #define CP_UMOUNT 0x00000001 195 #define CP_FASTBOOT 0x00000002 196 #define CP_SYNC 0x00000004 197 #define CP_RECOVERY 0x00000008 198 #define CP_DISCARD 0x00000010 199 #define CP_TRIMMED 0x00000020 200 #define CP_PAUSE 0x00000040 201 #define CP_RESIZE 0x00000080 202 203 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 204 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 205 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 206 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 207 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 208 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 209 #define DEF_CP_INTERVAL 60 /* 60 secs */ 210 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 211 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 212 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 213 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 214 215 struct cp_control { 216 int reason; 217 __u64 trim_start; 218 __u64 trim_end; 219 __u64 trim_minlen; 220 }; 221 222 /* 223 * indicate meta/data type 224 */ 225 enum { 226 META_CP, 227 META_NAT, 228 META_SIT, 229 META_SSA, 230 META_MAX, 231 META_POR, 232 DATA_GENERIC, /* check range only */ 233 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 234 DATA_GENERIC_ENHANCE_READ, /* 235 * strong check on range and segment 236 * bitmap but no warning due to race 237 * condition of read on truncated area 238 * by extent_cache 239 */ 240 META_GENERIC, 241 }; 242 243 /* for the list of ino */ 244 enum { 245 ORPHAN_INO, /* for orphan ino list */ 246 APPEND_INO, /* for append ino list */ 247 UPDATE_INO, /* for update ino list */ 248 TRANS_DIR_INO, /* for trasactions dir ino list */ 249 FLUSH_INO, /* for multiple device flushing */ 250 MAX_INO_ENTRY, /* max. list */ 251 }; 252 253 struct ino_entry { 254 struct list_head list; /* list head */ 255 nid_t ino; /* inode number */ 256 unsigned int dirty_device; /* dirty device bitmap */ 257 }; 258 259 /* for the list of inodes to be GCed */ 260 struct inode_entry { 261 struct list_head list; /* list head */ 262 struct inode *inode; /* vfs inode pointer */ 263 }; 264 265 struct fsync_node_entry { 266 struct list_head list; /* list head */ 267 struct page *page; /* warm node page pointer */ 268 unsigned int seq_id; /* sequence id */ 269 }; 270 271 struct ckpt_req { 272 struct completion wait; /* completion for checkpoint done */ 273 struct llist_node llnode; /* llist_node to be linked in wait queue */ 274 int ret; /* return code of checkpoint */ 275 ktime_t queue_time; /* request queued time */ 276 }; 277 278 struct ckpt_req_control { 279 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 280 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 281 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 282 atomic_t issued_ckpt; /* # of actually issued ckpts */ 283 atomic_t total_ckpt; /* # of total ckpts */ 284 atomic_t queued_ckpt; /* # of queued ckpts */ 285 struct llist_head issue_list; /* list for command issue */ 286 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 287 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 288 unsigned int peak_time; /* peak wait time in msec until now */ 289 }; 290 291 /* for the bitmap indicate blocks to be discarded */ 292 struct discard_entry { 293 struct list_head list; /* list head */ 294 block_t start_blkaddr; /* start blockaddr of current segment */ 295 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 296 }; 297 298 /* default discard granularity of inner discard thread, unit: block count */ 299 #define DEFAULT_DISCARD_GRANULARITY 16 300 301 /* max discard pend list number */ 302 #define MAX_PLIST_NUM 512 303 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 304 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 305 306 enum { 307 D_PREP, /* initial */ 308 D_PARTIAL, /* partially submitted */ 309 D_SUBMIT, /* all submitted */ 310 D_DONE, /* finished */ 311 }; 312 313 struct discard_info { 314 block_t lstart; /* logical start address */ 315 block_t len; /* length */ 316 block_t start; /* actual start address in dev */ 317 }; 318 319 struct discard_cmd { 320 struct rb_node rb_node; /* rb node located in rb-tree */ 321 union { 322 struct { 323 block_t lstart; /* logical start address */ 324 block_t len; /* length */ 325 block_t start; /* actual start address in dev */ 326 }; 327 struct discard_info di; /* discard info */ 328 329 }; 330 struct list_head list; /* command list */ 331 struct completion wait; /* compleation */ 332 struct block_device *bdev; /* bdev */ 333 unsigned short ref; /* reference count */ 334 unsigned char state; /* state */ 335 unsigned char queued; /* queued discard */ 336 int error; /* bio error */ 337 spinlock_t lock; /* for state/bio_ref updating */ 338 unsigned short bio_ref; /* bio reference count */ 339 }; 340 341 enum { 342 DPOLICY_BG, 343 DPOLICY_FORCE, 344 DPOLICY_FSTRIM, 345 DPOLICY_UMOUNT, 346 MAX_DPOLICY, 347 }; 348 349 struct discard_policy { 350 int type; /* type of discard */ 351 unsigned int min_interval; /* used for candidates exist */ 352 unsigned int mid_interval; /* used for device busy */ 353 unsigned int max_interval; /* used for candidates not exist */ 354 unsigned int max_requests; /* # of discards issued per round */ 355 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 356 bool io_aware; /* issue discard in idle time */ 357 bool sync; /* submit discard with REQ_SYNC flag */ 358 bool ordered; /* issue discard by lba order */ 359 bool timeout; /* discard timeout for put_super */ 360 unsigned int granularity; /* discard granularity */ 361 }; 362 363 struct discard_cmd_control { 364 struct task_struct *f2fs_issue_discard; /* discard thread */ 365 struct list_head entry_list; /* 4KB discard entry list */ 366 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 367 struct list_head wait_list; /* store on-flushing entries */ 368 struct list_head fstrim_list; /* in-flight discard from fstrim */ 369 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 370 unsigned int discard_wake; /* to wake up discard thread */ 371 struct mutex cmd_lock; 372 unsigned int nr_discards; /* # of discards in the list */ 373 unsigned int max_discards; /* max. discards to be issued */ 374 unsigned int discard_granularity; /* discard granularity */ 375 unsigned int undiscard_blks; /* # of undiscard blocks */ 376 unsigned int next_pos; /* next discard position */ 377 atomic_t issued_discard; /* # of issued discard */ 378 atomic_t queued_discard; /* # of queued discard */ 379 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 380 struct rb_root_cached root; /* root of discard rb-tree */ 381 bool rbtree_check; /* config for consistence check */ 382 }; 383 384 /* for the list of fsync inodes, used only during recovery */ 385 struct fsync_inode_entry { 386 struct list_head list; /* list head */ 387 struct inode *inode; /* vfs inode pointer */ 388 block_t blkaddr; /* block address locating the last fsync */ 389 block_t last_dentry; /* block address locating the last dentry */ 390 }; 391 392 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 393 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 394 395 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 396 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 397 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 398 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 399 400 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 401 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 402 403 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 404 { 405 int before = nats_in_cursum(journal); 406 407 journal->n_nats = cpu_to_le16(before + i); 408 return before; 409 } 410 411 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 412 { 413 int before = sits_in_cursum(journal); 414 415 journal->n_sits = cpu_to_le16(before + i); 416 return before; 417 } 418 419 static inline bool __has_cursum_space(struct f2fs_journal *journal, 420 int size, int type) 421 { 422 if (type == NAT_JOURNAL) 423 return size <= MAX_NAT_JENTRIES(journal); 424 return size <= MAX_SIT_JENTRIES(journal); 425 } 426 427 /* for inline stuff */ 428 #define DEF_INLINE_RESERVED_SIZE 1 429 static inline int get_extra_isize(struct inode *inode); 430 static inline int get_inline_xattr_addrs(struct inode *inode); 431 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 432 (CUR_ADDRS_PER_INODE(inode) - \ 433 get_inline_xattr_addrs(inode) - \ 434 DEF_INLINE_RESERVED_SIZE)) 435 436 /* for inline dir */ 437 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 438 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 439 BITS_PER_BYTE + 1)) 440 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 441 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 442 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 443 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 444 NR_INLINE_DENTRY(inode) + \ 445 INLINE_DENTRY_BITMAP_SIZE(inode))) 446 447 /* 448 * For INODE and NODE manager 449 */ 450 /* for directory operations */ 451 452 struct f2fs_filename { 453 /* 454 * The filename the user specified. This is NULL for some 455 * filesystem-internal operations, e.g. converting an inline directory 456 * to a non-inline one, or roll-forward recovering an encrypted dentry. 457 */ 458 const struct qstr *usr_fname; 459 460 /* 461 * The on-disk filename. For encrypted directories, this is encrypted. 462 * This may be NULL for lookups in an encrypted dir without the key. 463 */ 464 struct fscrypt_str disk_name; 465 466 /* The dirhash of this filename */ 467 f2fs_hash_t hash; 468 469 #ifdef CONFIG_FS_ENCRYPTION 470 /* 471 * For lookups in encrypted directories: either the buffer backing 472 * disk_name, or a buffer that holds the decoded no-key name. 473 */ 474 struct fscrypt_str crypto_buf; 475 #endif 476 #ifdef CONFIG_UNICODE 477 /* 478 * For casefolded directories: the casefolded name, but it's left NULL 479 * if the original name is not valid Unicode, if the directory is both 480 * casefolded and encrypted and its encryption key is unavailable, or if 481 * the filesystem is doing an internal operation where usr_fname is also 482 * NULL. In all these cases we fall back to treating the name as an 483 * opaque byte sequence. 484 */ 485 struct fscrypt_str cf_name; 486 #endif 487 }; 488 489 struct f2fs_dentry_ptr { 490 struct inode *inode; 491 void *bitmap; 492 struct f2fs_dir_entry *dentry; 493 __u8 (*filename)[F2FS_SLOT_LEN]; 494 int max; 495 int nr_bitmap; 496 }; 497 498 static inline void make_dentry_ptr_block(struct inode *inode, 499 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 500 { 501 d->inode = inode; 502 d->max = NR_DENTRY_IN_BLOCK; 503 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 504 d->bitmap = t->dentry_bitmap; 505 d->dentry = t->dentry; 506 d->filename = t->filename; 507 } 508 509 static inline void make_dentry_ptr_inline(struct inode *inode, 510 struct f2fs_dentry_ptr *d, void *t) 511 { 512 int entry_cnt = NR_INLINE_DENTRY(inode); 513 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 514 int reserved_size = INLINE_RESERVED_SIZE(inode); 515 516 d->inode = inode; 517 d->max = entry_cnt; 518 d->nr_bitmap = bitmap_size; 519 d->bitmap = t; 520 d->dentry = t + bitmap_size + reserved_size; 521 d->filename = t + bitmap_size + reserved_size + 522 SIZE_OF_DIR_ENTRY * entry_cnt; 523 } 524 525 /* 526 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 527 * as its node offset to distinguish from index node blocks. 528 * But some bits are used to mark the node block. 529 */ 530 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 531 >> OFFSET_BIT_SHIFT) 532 enum { 533 ALLOC_NODE, /* allocate a new node page if needed */ 534 LOOKUP_NODE, /* look up a node without readahead */ 535 LOOKUP_NODE_RA, /* 536 * look up a node with readahead called 537 * by get_data_block. 538 */ 539 }; 540 541 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */ 542 543 /* congestion wait timeout value, default: 20ms */ 544 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 545 546 /* maximum retry quota flush count */ 547 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 548 549 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 550 551 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 552 553 /* for in-memory extent cache entry */ 554 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 555 556 /* number of extent info in extent cache we try to shrink */ 557 #define EXTENT_CACHE_SHRINK_NUMBER 128 558 559 struct rb_entry { 560 struct rb_node rb_node; /* rb node located in rb-tree */ 561 union { 562 struct { 563 unsigned int ofs; /* start offset of the entry */ 564 unsigned int len; /* length of the entry */ 565 }; 566 unsigned long long key; /* 64-bits key */ 567 } __packed; 568 }; 569 570 struct extent_info { 571 unsigned int fofs; /* start offset in a file */ 572 unsigned int len; /* length of the extent */ 573 u32 blk; /* start block address of the extent */ 574 }; 575 576 struct extent_node { 577 struct rb_node rb_node; /* rb node located in rb-tree */ 578 struct extent_info ei; /* extent info */ 579 struct list_head list; /* node in global extent list of sbi */ 580 struct extent_tree *et; /* extent tree pointer */ 581 }; 582 583 struct extent_tree { 584 nid_t ino; /* inode number */ 585 struct rb_root_cached root; /* root of extent info rb-tree */ 586 struct extent_node *cached_en; /* recently accessed extent node */ 587 struct extent_info largest; /* largested extent info */ 588 struct list_head list; /* to be used by sbi->zombie_list */ 589 rwlock_t lock; /* protect extent info rb-tree */ 590 atomic_t node_cnt; /* # of extent node in rb-tree*/ 591 bool largest_updated; /* largest extent updated */ 592 }; 593 594 /* 595 * This structure is taken from ext4_map_blocks. 596 * 597 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 598 */ 599 #define F2FS_MAP_NEW (1 << BH_New) 600 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 601 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 602 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 603 F2FS_MAP_UNWRITTEN) 604 605 struct f2fs_map_blocks { 606 block_t m_pblk; 607 block_t m_lblk; 608 unsigned int m_len; 609 unsigned int m_flags; 610 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 611 pgoff_t *m_next_extent; /* point to next possible extent */ 612 int m_seg_type; 613 bool m_may_create; /* indicate it is from write path */ 614 }; 615 616 /* for flag in get_data_block */ 617 enum { 618 F2FS_GET_BLOCK_DEFAULT, 619 F2FS_GET_BLOCK_FIEMAP, 620 F2FS_GET_BLOCK_BMAP, 621 F2FS_GET_BLOCK_DIO, 622 F2FS_GET_BLOCK_PRE_DIO, 623 F2FS_GET_BLOCK_PRE_AIO, 624 F2FS_GET_BLOCK_PRECACHE, 625 }; 626 627 /* 628 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 629 */ 630 #define FADVISE_COLD_BIT 0x01 631 #define FADVISE_LOST_PINO_BIT 0x02 632 #define FADVISE_ENCRYPT_BIT 0x04 633 #define FADVISE_ENC_NAME_BIT 0x08 634 #define FADVISE_KEEP_SIZE_BIT 0x10 635 #define FADVISE_HOT_BIT 0x20 636 #define FADVISE_VERITY_BIT 0x40 637 638 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 639 640 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 641 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 642 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 643 644 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 645 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 646 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 647 648 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 649 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 650 651 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 652 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 653 654 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 655 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 656 657 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 658 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 659 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 660 661 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 662 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 663 664 #define DEF_DIR_LEVEL 0 665 666 enum { 667 GC_FAILURE_PIN, 668 GC_FAILURE_ATOMIC, 669 MAX_GC_FAILURE 670 }; 671 672 /* used for f2fs_inode_info->flags */ 673 enum { 674 FI_NEW_INODE, /* indicate newly allocated inode */ 675 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 676 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 677 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 678 FI_INC_LINK, /* need to increment i_nlink */ 679 FI_ACL_MODE, /* indicate acl mode */ 680 FI_NO_ALLOC, /* should not allocate any blocks */ 681 FI_FREE_NID, /* free allocated nide */ 682 FI_NO_EXTENT, /* not to use the extent cache */ 683 FI_INLINE_XATTR, /* used for inline xattr */ 684 FI_INLINE_DATA, /* used for inline data*/ 685 FI_INLINE_DENTRY, /* used for inline dentry */ 686 FI_APPEND_WRITE, /* inode has appended data */ 687 FI_UPDATE_WRITE, /* inode has in-place-update data */ 688 FI_NEED_IPU, /* used for ipu per file */ 689 FI_ATOMIC_FILE, /* indicate atomic file */ 690 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 691 FI_VOLATILE_FILE, /* indicate volatile file */ 692 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 693 FI_DROP_CACHE, /* drop dirty page cache */ 694 FI_DATA_EXIST, /* indicate data exists */ 695 FI_INLINE_DOTS, /* indicate inline dot dentries */ 696 FI_DO_DEFRAG, /* indicate defragment is running */ 697 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 698 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 699 FI_HOT_DATA, /* indicate file is hot */ 700 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 701 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 702 FI_PIN_FILE, /* indicate file should not be gced */ 703 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 704 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 705 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 706 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 707 FI_MMAP_FILE, /* indicate file was mmapped */ 708 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 709 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 710 FI_MAX, /* max flag, never be used */ 711 }; 712 713 struct f2fs_inode_info { 714 struct inode vfs_inode; /* serve a vfs inode */ 715 unsigned long i_flags; /* keep an inode flags for ioctl */ 716 unsigned char i_advise; /* use to give file attribute hints */ 717 unsigned char i_dir_level; /* use for dentry level for large dir */ 718 unsigned int i_current_depth; /* only for directory depth */ 719 /* for gc failure statistic */ 720 unsigned int i_gc_failures[MAX_GC_FAILURE]; 721 unsigned int i_pino; /* parent inode number */ 722 umode_t i_acl_mode; /* keep file acl mode temporarily */ 723 724 /* Use below internally in f2fs*/ 725 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 726 struct rw_semaphore i_sem; /* protect fi info */ 727 atomic_t dirty_pages; /* # of dirty pages */ 728 f2fs_hash_t chash; /* hash value of given file name */ 729 unsigned int clevel; /* maximum level of given file name */ 730 struct task_struct *task; /* lookup and create consistency */ 731 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 732 nid_t i_xattr_nid; /* node id that contains xattrs */ 733 loff_t last_disk_size; /* lastly written file size */ 734 spinlock_t i_size_lock; /* protect last_disk_size */ 735 736 #ifdef CONFIG_QUOTA 737 struct dquot *i_dquot[MAXQUOTAS]; 738 739 /* quota space reservation, managed internally by quota code */ 740 qsize_t i_reserved_quota; 741 #endif 742 struct list_head dirty_list; /* dirty list for dirs and files */ 743 struct list_head gdirty_list; /* linked in global dirty list */ 744 struct list_head inmem_ilist; /* list for inmem inodes */ 745 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 746 struct task_struct *inmem_task; /* store inmemory task */ 747 struct mutex inmem_lock; /* lock for inmemory pages */ 748 struct extent_tree *extent_tree; /* cached extent_tree entry */ 749 750 /* avoid racing between foreground op and gc */ 751 struct rw_semaphore i_gc_rwsem[2]; 752 struct rw_semaphore i_mmap_sem; 753 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 754 755 int i_extra_isize; /* size of extra space located in i_addr */ 756 kprojid_t i_projid; /* id for project quota */ 757 int i_inline_xattr_size; /* inline xattr size */ 758 struct timespec64 i_crtime; /* inode creation time */ 759 struct timespec64 i_disk_time[4];/* inode disk times */ 760 761 /* for file compress */ 762 atomic_t i_compr_blocks; /* # of compressed blocks */ 763 unsigned char i_compress_algorithm; /* algorithm type */ 764 unsigned char i_log_cluster_size; /* log of cluster size */ 765 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 766 unsigned short i_compress_flag; /* compress flag */ 767 unsigned int i_cluster_size; /* cluster size */ 768 }; 769 770 static inline void get_extent_info(struct extent_info *ext, 771 struct f2fs_extent *i_ext) 772 { 773 ext->fofs = le32_to_cpu(i_ext->fofs); 774 ext->blk = le32_to_cpu(i_ext->blk); 775 ext->len = le32_to_cpu(i_ext->len); 776 } 777 778 static inline void set_raw_extent(struct extent_info *ext, 779 struct f2fs_extent *i_ext) 780 { 781 i_ext->fofs = cpu_to_le32(ext->fofs); 782 i_ext->blk = cpu_to_le32(ext->blk); 783 i_ext->len = cpu_to_le32(ext->len); 784 } 785 786 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 787 u32 blk, unsigned int len) 788 { 789 ei->fofs = fofs; 790 ei->blk = blk; 791 ei->len = len; 792 } 793 794 static inline bool __is_discard_mergeable(struct discard_info *back, 795 struct discard_info *front, unsigned int max_len) 796 { 797 return (back->lstart + back->len == front->lstart) && 798 (back->len + front->len <= max_len); 799 } 800 801 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 802 struct discard_info *back, unsigned int max_len) 803 { 804 return __is_discard_mergeable(back, cur, max_len); 805 } 806 807 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 808 struct discard_info *front, unsigned int max_len) 809 { 810 return __is_discard_mergeable(cur, front, max_len); 811 } 812 813 static inline bool __is_extent_mergeable(struct extent_info *back, 814 struct extent_info *front) 815 { 816 return (back->fofs + back->len == front->fofs && 817 back->blk + back->len == front->blk); 818 } 819 820 static inline bool __is_back_mergeable(struct extent_info *cur, 821 struct extent_info *back) 822 { 823 return __is_extent_mergeable(back, cur); 824 } 825 826 static inline bool __is_front_mergeable(struct extent_info *cur, 827 struct extent_info *front) 828 { 829 return __is_extent_mergeable(cur, front); 830 } 831 832 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 833 static inline void __try_update_largest_extent(struct extent_tree *et, 834 struct extent_node *en) 835 { 836 if (en->ei.len > et->largest.len) { 837 et->largest = en->ei; 838 et->largest_updated = true; 839 } 840 } 841 842 /* 843 * For free nid management 844 */ 845 enum nid_state { 846 FREE_NID, /* newly added to free nid list */ 847 PREALLOC_NID, /* it is preallocated */ 848 MAX_NID_STATE, 849 }; 850 851 enum nat_state { 852 TOTAL_NAT, 853 DIRTY_NAT, 854 RECLAIMABLE_NAT, 855 MAX_NAT_STATE, 856 }; 857 858 struct f2fs_nm_info { 859 block_t nat_blkaddr; /* base disk address of NAT */ 860 nid_t max_nid; /* maximum possible node ids */ 861 nid_t available_nids; /* # of available node ids */ 862 nid_t next_scan_nid; /* the next nid to be scanned */ 863 unsigned int ram_thresh; /* control the memory footprint */ 864 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 865 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 866 867 /* NAT cache management */ 868 struct radix_tree_root nat_root;/* root of the nat entry cache */ 869 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 870 struct rw_semaphore nat_tree_lock; /* protect nat entry tree */ 871 struct list_head nat_entries; /* cached nat entry list (clean) */ 872 spinlock_t nat_list_lock; /* protect clean nat entry list */ 873 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 874 unsigned int nat_blocks; /* # of nat blocks */ 875 876 /* free node ids management */ 877 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 878 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 879 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 880 spinlock_t nid_list_lock; /* protect nid lists ops */ 881 struct mutex build_lock; /* lock for build free nids */ 882 unsigned char **free_nid_bitmap; 883 unsigned char *nat_block_bitmap; 884 unsigned short *free_nid_count; /* free nid count of NAT block */ 885 886 /* for checkpoint */ 887 char *nat_bitmap; /* NAT bitmap pointer */ 888 889 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 890 unsigned char *nat_bits; /* NAT bits blocks */ 891 unsigned char *full_nat_bits; /* full NAT pages */ 892 unsigned char *empty_nat_bits; /* empty NAT pages */ 893 #ifdef CONFIG_F2FS_CHECK_FS 894 char *nat_bitmap_mir; /* NAT bitmap mirror */ 895 #endif 896 int bitmap_size; /* bitmap size */ 897 }; 898 899 /* 900 * this structure is used as one of function parameters. 901 * all the information are dedicated to a given direct node block determined 902 * by the data offset in a file. 903 */ 904 struct dnode_of_data { 905 struct inode *inode; /* vfs inode pointer */ 906 struct page *inode_page; /* its inode page, NULL is possible */ 907 struct page *node_page; /* cached direct node page */ 908 nid_t nid; /* node id of the direct node block */ 909 unsigned int ofs_in_node; /* data offset in the node page */ 910 bool inode_page_locked; /* inode page is locked or not */ 911 bool node_changed; /* is node block changed */ 912 char cur_level; /* level of hole node page */ 913 char max_level; /* level of current page located */ 914 block_t data_blkaddr; /* block address of the node block */ 915 }; 916 917 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 918 struct page *ipage, struct page *npage, nid_t nid) 919 { 920 memset(dn, 0, sizeof(*dn)); 921 dn->inode = inode; 922 dn->inode_page = ipage; 923 dn->node_page = npage; 924 dn->nid = nid; 925 } 926 927 /* 928 * For SIT manager 929 * 930 * By default, there are 6 active log areas across the whole main area. 931 * When considering hot and cold data separation to reduce cleaning overhead, 932 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 933 * respectively. 934 * In the current design, you should not change the numbers intentionally. 935 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 936 * logs individually according to the underlying devices. (default: 6) 937 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 938 * data and 8 for node logs. 939 */ 940 #define NR_CURSEG_DATA_TYPE (3) 941 #define NR_CURSEG_NODE_TYPE (3) 942 #define NR_CURSEG_INMEM_TYPE (2) 943 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 944 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 945 946 enum { 947 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 948 CURSEG_WARM_DATA, /* data blocks */ 949 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 950 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 951 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 952 CURSEG_COLD_NODE, /* indirect node blocks */ 953 NR_PERSISTENT_LOG, /* number of persistent log */ 954 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 955 /* pinned file that needs consecutive block address */ 956 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 957 NO_CHECK_TYPE, /* number of persistent & inmem log */ 958 }; 959 960 struct flush_cmd { 961 struct completion wait; 962 struct llist_node llnode; 963 nid_t ino; 964 int ret; 965 }; 966 967 struct flush_cmd_control { 968 struct task_struct *f2fs_issue_flush; /* flush thread */ 969 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 970 atomic_t issued_flush; /* # of issued flushes */ 971 atomic_t queued_flush; /* # of queued flushes */ 972 struct llist_head issue_list; /* list for command issue */ 973 struct llist_node *dispatch_list; /* list for command dispatch */ 974 }; 975 976 struct f2fs_sm_info { 977 struct sit_info *sit_info; /* whole segment information */ 978 struct free_segmap_info *free_info; /* free segment information */ 979 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 980 struct curseg_info *curseg_array; /* active segment information */ 981 982 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 983 984 block_t seg0_blkaddr; /* block address of 0'th segment */ 985 block_t main_blkaddr; /* start block address of main area */ 986 block_t ssa_blkaddr; /* start block address of SSA area */ 987 988 unsigned int segment_count; /* total # of segments */ 989 unsigned int main_segments; /* # of segments in main area */ 990 unsigned int reserved_segments; /* # of reserved segments */ 991 unsigned int ovp_segments; /* # of overprovision segments */ 992 993 /* a threshold to reclaim prefree segments */ 994 unsigned int rec_prefree_segments; 995 996 /* for batched trimming */ 997 unsigned int trim_sections; /* # of sections to trim */ 998 999 struct list_head sit_entry_set; /* sit entry set list */ 1000 1001 unsigned int ipu_policy; /* in-place-update policy */ 1002 unsigned int min_ipu_util; /* in-place-update threshold */ 1003 unsigned int min_fsync_blocks; /* threshold for fsync */ 1004 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1005 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1006 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1007 1008 /* for flush command control */ 1009 struct flush_cmd_control *fcc_info; 1010 1011 /* for discard command control */ 1012 struct discard_cmd_control *dcc_info; 1013 }; 1014 1015 /* 1016 * For superblock 1017 */ 1018 /* 1019 * COUNT_TYPE for monitoring 1020 * 1021 * f2fs monitors the number of several block types such as on-writeback, 1022 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1023 */ 1024 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1025 enum count_type { 1026 F2FS_DIRTY_DENTS, 1027 F2FS_DIRTY_DATA, 1028 F2FS_DIRTY_QDATA, 1029 F2FS_DIRTY_NODES, 1030 F2FS_DIRTY_META, 1031 F2FS_INMEM_PAGES, 1032 F2FS_DIRTY_IMETA, 1033 F2FS_WB_CP_DATA, 1034 F2FS_WB_DATA, 1035 F2FS_RD_DATA, 1036 F2FS_RD_NODE, 1037 F2FS_RD_META, 1038 F2FS_DIO_WRITE, 1039 F2FS_DIO_READ, 1040 NR_COUNT_TYPE, 1041 }; 1042 1043 /* 1044 * The below are the page types of bios used in submit_bio(). 1045 * The available types are: 1046 * DATA User data pages. It operates as async mode. 1047 * NODE Node pages. It operates as async mode. 1048 * META FS metadata pages such as SIT, NAT, CP. 1049 * NR_PAGE_TYPE The number of page types. 1050 * META_FLUSH Make sure the previous pages are written 1051 * with waiting the bio's completion 1052 * ... Only can be used with META. 1053 */ 1054 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1055 enum page_type { 1056 DATA, 1057 NODE, 1058 META, 1059 NR_PAGE_TYPE, 1060 META_FLUSH, 1061 INMEM, /* the below types are used by tracepoints only. */ 1062 INMEM_DROP, 1063 INMEM_INVALIDATE, 1064 INMEM_REVOKE, 1065 IPU, 1066 OPU, 1067 }; 1068 1069 enum temp_type { 1070 HOT = 0, /* must be zero for meta bio */ 1071 WARM, 1072 COLD, 1073 NR_TEMP_TYPE, 1074 }; 1075 1076 enum need_lock_type { 1077 LOCK_REQ = 0, 1078 LOCK_DONE, 1079 LOCK_RETRY, 1080 }; 1081 1082 enum cp_reason_type { 1083 CP_NO_NEEDED, 1084 CP_NON_REGULAR, 1085 CP_COMPRESSED, 1086 CP_HARDLINK, 1087 CP_SB_NEED_CP, 1088 CP_WRONG_PINO, 1089 CP_NO_SPC_ROLL, 1090 CP_NODE_NEED_CP, 1091 CP_FASTBOOT_MODE, 1092 CP_SPEC_LOG_NUM, 1093 CP_RECOVER_DIR, 1094 }; 1095 1096 enum iostat_type { 1097 /* WRITE IO */ 1098 APP_DIRECT_IO, /* app direct write IOs */ 1099 APP_BUFFERED_IO, /* app buffered write IOs */ 1100 APP_WRITE_IO, /* app write IOs */ 1101 APP_MAPPED_IO, /* app mapped IOs */ 1102 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1103 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1104 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1105 FS_GC_DATA_IO, /* data IOs from forground gc */ 1106 FS_GC_NODE_IO, /* node IOs from forground gc */ 1107 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1108 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1109 FS_CP_META_IO, /* meta IOs from checkpoint */ 1110 1111 /* READ IO */ 1112 APP_DIRECT_READ_IO, /* app direct read IOs */ 1113 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1114 APP_READ_IO, /* app read IOs */ 1115 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1116 FS_DATA_READ_IO, /* data read IOs */ 1117 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1118 FS_CDATA_READ_IO, /* compressed data read IOs */ 1119 FS_NODE_READ_IO, /* node read IOs */ 1120 FS_META_READ_IO, /* meta read IOs */ 1121 1122 /* other */ 1123 FS_DISCARD, /* discard */ 1124 NR_IO_TYPE, 1125 }; 1126 1127 struct f2fs_io_info { 1128 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1129 nid_t ino; /* inode number */ 1130 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1131 enum temp_type temp; /* contains HOT/WARM/COLD */ 1132 int op; /* contains REQ_OP_ */ 1133 int op_flags; /* req_flag_bits */ 1134 block_t new_blkaddr; /* new block address to be written */ 1135 block_t old_blkaddr; /* old block address before Cow */ 1136 struct page *page; /* page to be written */ 1137 struct page *encrypted_page; /* encrypted page */ 1138 struct page *compressed_page; /* compressed page */ 1139 struct list_head list; /* serialize IOs */ 1140 bool submitted; /* indicate IO submission */ 1141 int need_lock; /* indicate we need to lock cp_rwsem */ 1142 bool in_list; /* indicate fio is in io_list */ 1143 bool is_por; /* indicate IO is from recovery or not */ 1144 bool retry; /* need to reallocate block address */ 1145 int compr_blocks; /* # of compressed block addresses */ 1146 bool encrypted; /* indicate file is encrypted */ 1147 enum iostat_type io_type; /* io type */ 1148 struct writeback_control *io_wbc; /* writeback control */ 1149 struct bio **bio; /* bio for ipu */ 1150 sector_t *last_block; /* last block number in bio */ 1151 unsigned char version; /* version of the node */ 1152 }; 1153 1154 struct bio_entry { 1155 struct bio *bio; 1156 struct list_head list; 1157 }; 1158 1159 #define is_read_io(rw) ((rw) == READ) 1160 struct f2fs_bio_info { 1161 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1162 struct bio *bio; /* bios to merge */ 1163 sector_t last_block_in_bio; /* last block number */ 1164 struct f2fs_io_info fio; /* store buffered io info. */ 1165 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1166 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1167 struct list_head io_list; /* track fios */ 1168 struct list_head bio_list; /* bio entry list head */ 1169 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1170 }; 1171 1172 #define FDEV(i) (sbi->devs[i]) 1173 #define RDEV(i) (raw_super->devs[i]) 1174 struct f2fs_dev_info { 1175 struct block_device *bdev; 1176 char path[MAX_PATH_LEN]; 1177 unsigned int total_segments; 1178 block_t start_blk; 1179 block_t end_blk; 1180 #ifdef CONFIG_BLK_DEV_ZONED 1181 unsigned int nr_blkz; /* Total number of zones */ 1182 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1183 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1184 #endif 1185 }; 1186 1187 enum inode_type { 1188 DIR_INODE, /* for dirty dir inode */ 1189 FILE_INODE, /* for dirty regular/symlink inode */ 1190 DIRTY_META, /* for all dirtied inode metadata */ 1191 ATOMIC_FILE, /* for all atomic files */ 1192 NR_INODE_TYPE, 1193 }; 1194 1195 /* for inner inode cache management */ 1196 struct inode_management { 1197 struct radix_tree_root ino_root; /* ino entry array */ 1198 spinlock_t ino_lock; /* for ino entry lock */ 1199 struct list_head ino_list; /* inode list head */ 1200 unsigned long ino_num; /* number of entries */ 1201 }; 1202 1203 /* for GC_AT */ 1204 struct atgc_management { 1205 bool atgc_enabled; /* ATGC is enabled or not */ 1206 struct rb_root_cached root; /* root of victim rb-tree */ 1207 struct list_head victim_list; /* linked with all victim entries */ 1208 unsigned int victim_count; /* victim count in rb-tree */ 1209 unsigned int candidate_ratio; /* candidate ratio */ 1210 unsigned int max_candidate_count; /* max candidate count */ 1211 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1212 unsigned long long age_threshold; /* age threshold */ 1213 }; 1214 1215 /* For s_flag in struct f2fs_sb_info */ 1216 enum { 1217 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1218 SBI_IS_CLOSE, /* specify unmounting */ 1219 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1220 SBI_POR_DOING, /* recovery is doing or not */ 1221 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1222 SBI_NEED_CP, /* need to checkpoint */ 1223 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1224 SBI_IS_RECOVERED, /* recovered orphan/data */ 1225 SBI_CP_DISABLED, /* CP was disabled last mount */ 1226 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1227 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1228 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1229 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1230 SBI_IS_RESIZEFS, /* resizefs is in process */ 1231 }; 1232 1233 enum { 1234 CP_TIME, 1235 REQ_TIME, 1236 DISCARD_TIME, 1237 GC_TIME, 1238 DISABLE_TIME, 1239 UMOUNT_DISCARD_TIMEOUT, 1240 MAX_TIME, 1241 }; 1242 1243 enum { 1244 GC_NORMAL, 1245 GC_IDLE_CB, 1246 GC_IDLE_GREEDY, 1247 GC_IDLE_AT, 1248 GC_URGENT_HIGH, 1249 GC_URGENT_LOW, 1250 }; 1251 1252 enum { 1253 BGGC_MODE_ON, /* background gc is on */ 1254 BGGC_MODE_OFF, /* background gc is off */ 1255 BGGC_MODE_SYNC, /* 1256 * background gc is on, migrating blocks 1257 * like foreground gc 1258 */ 1259 }; 1260 1261 enum { 1262 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1263 FS_MODE_LFS, /* use lfs allocation only */ 1264 }; 1265 1266 enum { 1267 WHINT_MODE_OFF, /* not pass down write hints */ 1268 WHINT_MODE_USER, /* try to pass down hints given by users */ 1269 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1270 }; 1271 1272 enum { 1273 ALLOC_MODE_DEFAULT, /* stay default */ 1274 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1275 }; 1276 1277 enum fsync_mode { 1278 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1279 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1280 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1281 }; 1282 1283 enum { 1284 COMPR_MODE_FS, /* 1285 * automatically compress compression 1286 * enabled files 1287 */ 1288 COMPR_MODE_USER, /* 1289 * automatical compression is disabled. 1290 * user can control the file compression 1291 * using ioctls 1292 */ 1293 }; 1294 1295 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1296 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1297 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1298 1299 /* 1300 * Layout of f2fs page.private: 1301 * 1302 * Layout A: lowest bit should be 1 1303 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1304 * bit 0 PAGE_PRIVATE_NOT_POINTER 1305 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1306 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1307 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1308 * bit 4 PAGE_PRIVATE_INLINE_INODE 1309 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1310 * bit 6- f2fs private data 1311 * 1312 * Layout B: lowest bit should be 0 1313 * page.private is a wrapped pointer. 1314 */ 1315 enum { 1316 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1317 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1318 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1319 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1320 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1321 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1322 PAGE_PRIVATE_MAX 1323 }; 1324 1325 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1326 static inline bool page_private_##name(struct page *page) \ 1327 { \ 1328 return test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1329 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1330 } 1331 1332 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1333 static inline void set_page_private_##name(struct page *page) \ 1334 { \ 1335 if (!PagePrivate(page)) { \ 1336 get_page(page); \ 1337 SetPagePrivate(page); \ 1338 } \ 1339 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1340 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1341 } 1342 1343 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1344 static inline void clear_page_private_##name(struct page *page) \ 1345 { \ 1346 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1347 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1348 set_page_private(page, 0); \ 1349 if (PagePrivate(page)) { \ 1350 ClearPagePrivate(page); \ 1351 put_page(page); \ 1352 }\ 1353 } \ 1354 } 1355 1356 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1357 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1358 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1359 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1360 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1361 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1362 1363 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1364 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1365 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1366 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1367 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1368 1369 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1370 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1371 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1372 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1373 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1374 1375 /* For compression */ 1376 enum compress_algorithm_type { 1377 COMPRESS_LZO, 1378 COMPRESS_LZ4, 1379 COMPRESS_ZSTD, 1380 COMPRESS_LZORLE, 1381 COMPRESS_MAX, 1382 }; 1383 1384 enum compress_flag { 1385 COMPRESS_CHKSUM, 1386 COMPRESS_MAX_FLAG, 1387 }; 1388 1389 #define COMPRESS_DATA_RESERVED_SIZE 4 1390 struct compress_data { 1391 __le32 clen; /* compressed data size */ 1392 __le32 chksum; /* compressed data chksum */ 1393 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1394 u8 cdata[]; /* compressed data */ 1395 }; 1396 1397 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1398 1399 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1400 1401 #define COMPRESS_LEVEL_OFFSET 8 1402 1403 /* compress context */ 1404 struct compress_ctx { 1405 struct inode *inode; /* inode the context belong to */ 1406 pgoff_t cluster_idx; /* cluster index number */ 1407 unsigned int cluster_size; /* page count in cluster */ 1408 unsigned int log_cluster_size; /* log of cluster size */ 1409 struct page **rpages; /* pages store raw data in cluster */ 1410 unsigned int nr_rpages; /* total page number in rpages */ 1411 struct page **cpages; /* pages store compressed data in cluster */ 1412 unsigned int nr_cpages; /* total page number in cpages */ 1413 void *rbuf; /* virtual mapped address on rpages */ 1414 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1415 size_t rlen; /* valid data length in rbuf */ 1416 size_t clen; /* valid data length in cbuf */ 1417 void *private; /* payload buffer for specified compression algorithm */ 1418 void *private2; /* extra payload buffer */ 1419 }; 1420 1421 /* compress context for write IO path */ 1422 struct compress_io_ctx { 1423 u32 magic; /* magic number to indicate page is compressed */ 1424 struct inode *inode; /* inode the context belong to */ 1425 struct page **rpages; /* pages store raw data in cluster */ 1426 unsigned int nr_rpages; /* total page number in rpages */ 1427 atomic_t pending_pages; /* in-flight compressed page count */ 1428 }; 1429 1430 /* Context for decompressing one cluster on the read IO path */ 1431 struct decompress_io_ctx { 1432 u32 magic; /* magic number to indicate page is compressed */ 1433 struct inode *inode; /* inode the context belong to */ 1434 pgoff_t cluster_idx; /* cluster index number */ 1435 unsigned int cluster_size; /* page count in cluster */ 1436 unsigned int log_cluster_size; /* log of cluster size */ 1437 struct page **rpages; /* pages store raw data in cluster */ 1438 unsigned int nr_rpages; /* total page number in rpages */ 1439 struct page **cpages; /* pages store compressed data in cluster */ 1440 unsigned int nr_cpages; /* total page number in cpages */ 1441 struct page **tpages; /* temp pages to pad holes in cluster */ 1442 void *rbuf; /* virtual mapped address on rpages */ 1443 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1444 size_t rlen; /* valid data length in rbuf */ 1445 size_t clen; /* valid data length in cbuf */ 1446 1447 /* 1448 * The number of compressed pages remaining to be read in this cluster. 1449 * This is initially nr_cpages. It is decremented by 1 each time a page 1450 * has been read (or failed to be read). When it reaches 0, the cluster 1451 * is decompressed (or an error is reported). 1452 * 1453 * If an error occurs before all the pages have been submitted for I/O, 1454 * then this will never reach 0. In this case the I/O submitter is 1455 * responsible for calling f2fs_decompress_end_io() instead. 1456 */ 1457 atomic_t remaining_pages; 1458 1459 /* 1460 * Number of references to this decompress_io_ctx. 1461 * 1462 * One reference is held for I/O completion. This reference is dropped 1463 * after the pagecache pages are updated and unlocked -- either after 1464 * decompression (and verity if enabled), or after an error. 1465 * 1466 * In addition, each compressed page holds a reference while it is in a 1467 * bio. These references are necessary prevent compressed pages from 1468 * being freed while they are still in a bio. 1469 */ 1470 refcount_t refcnt; 1471 1472 bool failed; /* IO error occurred before decompression? */ 1473 bool need_verity; /* need fs-verity verification after decompression? */ 1474 void *private; /* payload buffer for specified decompression algorithm */ 1475 void *private2; /* extra payload buffer */ 1476 struct work_struct verity_work; /* work to verify the decompressed pages */ 1477 }; 1478 1479 #define NULL_CLUSTER ((unsigned int)(~0)) 1480 #define MIN_COMPRESS_LOG_SIZE 2 1481 #define MAX_COMPRESS_LOG_SIZE 8 1482 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1483 1484 struct f2fs_sb_info { 1485 struct super_block *sb; /* pointer to VFS super block */ 1486 struct proc_dir_entry *s_proc; /* proc entry */ 1487 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1488 struct rw_semaphore sb_lock; /* lock for raw super block */ 1489 int valid_super_block; /* valid super block no */ 1490 unsigned long s_flag; /* flags for sbi */ 1491 struct mutex writepages; /* mutex for writepages() */ 1492 1493 #ifdef CONFIG_BLK_DEV_ZONED 1494 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1495 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1496 #endif 1497 1498 /* for node-related operations */ 1499 struct f2fs_nm_info *nm_info; /* node manager */ 1500 struct inode *node_inode; /* cache node blocks */ 1501 1502 /* for segment-related operations */ 1503 struct f2fs_sm_info *sm_info; /* segment manager */ 1504 1505 /* for bio operations */ 1506 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1507 /* keep migration IO order for LFS mode */ 1508 struct rw_semaphore io_order_lock; 1509 mempool_t *write_io_dummy; /* Dummy pages */ 1510 1511 /* for checkpoint */ 1512 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1513 int cur_cp_pack; /* remain current cp pack */ 1514 spinlock_t cp_lock; /* for flag in ckpt */ 1515 struct inode *meta_inode; /* cache meta blocks */ 1516 struct rw_semaphore cp_global_sem; /* checkpoint procedure lock */ 1517 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1518 struct rw_semaphore node_write; /* locking node writes */ 1519 struct rw_semaphore node_change; /* locking node change */ 1520 wait_queue_head_t cp_wait; 1521 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1522 long interval_time[MAX_TIME]; /* to store thresholds */ 1523 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1524 1525 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1526 1527 spinlock_t fsync_node_lock; /* for node entry lock */ 1528 struct list_head fsync_node_list; /* node list head */ 1529 unsigned int fsync_seg_id; /* sequence id */ 1530 unsigned int fsync_node_num; /* number of node entries */ 1531 1532 /* for orphan inode, use 0'th array */ 1533 unsigned int max_orphans; /* max orphan inodes */ 1534 1535 /* for inode management */ 1536 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1537 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1538 struct mutex flush_lock; /* for flush exclusion */ 1539 1540 /* for extent tree cache */ 1541 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1542 struct mutex extent_tree_lock; /* locking extent radix tree */ 1543 struct list_head extent_list; /* lru list for shrinker */ 1544 spinlock_t extent_lock; /* locking extent lru list */ 1545 atomic_t total_ext_tree; /* extent tree count */ 1546 struct list_head zombie_list; /* extent zombie tree list */ 1547 atomic_t total_zombie_tree; /* extent zombie tree count */ 1548 atomic_t total_ext_node; /* extent info count */ 1549 1550 /* basic filesystem units */ 1551 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1552 unsigned int log_blocksize; /* log2 block size */ 1553 unsigned int blocksize; /* block size */ 1554 unsigned int root_ino_num; /* root inode number*/ 1555 unsigned int node_ino_num; /* node inode number*/ 1556 unsigned int meta_ino_num; /* meta inode number*/ 1557 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1558 unsigned int blocks_per_seg; /* blocks per segment */ 1559 unsigned int segs_per_sec; /* segments per section */ 1560 unsigned int secs_per_zone; /* sections per zone */ 1561 unsigned int total_sections; /* total section count */ 1562 unsigned int total_node_count; /* total node block count */ 1563 unsigned int total_valid_node_count; /* valid node block count */ 1564 int dir_level; /* directory level */ 1565 int readdir_ra; /* readahead inode in readdir */ 1566 u64 max_io_bytes; /* max io bytes to merge IOs */ 1567 1568 block_t user_block_count; /* # of user blocks */ 1569 block_t total_valid_block_count; /* # of valid blocks */ 1570 block_t discard_blks; /* discard command candidats */ 1571 block_t last_valid_block_count; /* for recovery */ 1572 block_t reserved_blocks; /* configurable reserved blocks */ 1573 block_t current_reserved_blocks; /* current reserved blocks */ 1574 1575 /* Additional tracking for no checkpoint mode */ 1576 block_t unusable_block_count; /* # of blocks saved by last cp */ 1577 1578 unsigned int nquota_files; /* # of quota sysfile */ 1579 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1580 1581 /* # of pages, see count_type */ 1582 atomic_t nr_pages[NR_COUNT_TYPE]; 1583 /* # of allocated blocks */ 1584 struct percpu_counter alloc_valid_block_count; 1585 1586 /* writeback control */ 1587 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1588 1589 /* valid inode count */ 1590 struct percpu_counter total_valid_inode_count; 1591 1592 struct f2fs_mount_info mount_opt; /* mount options */ 1593 1594 /* for cleaning operations */ 1595 struct rw_semaphore gc_lock; /* 1596 * semaphore for GC, avoid 1597 * race between GC and GC or CP 1598 */ 1599 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1600 struct atgc_management am; /* atgc management */ 1601 unsigned int cur_victim_sec; /* current victim section num */ 1602 unsigned int gc_mode; /* current GC state */ 1603 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1604 1605 /* for skip statistic */ 1606 unsigned int atomic_files; /* # of opened atomic file */ 1607 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1608 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1609 1610 /* threshold for gc trials on pinned files */ 1611 u64 gc_pin_file_threshold; 1612 struct rw_semaphore pin_sem; 1613 1614 /* maximum # of trials to find a victim segment for SSR and GC */ 1615 unsigned int max_victim_search; 1616 /* migration granularity of garbage collection, unit: segment */ 1617 unsigned int migration_granularity; 1618 1619 /* 1620 * for stat information. 1621 * one is for the LFS mode, and the other is for the SSR mode. 1622 */ 1623 #ifdef CONFIG_F2FS_STAT_FS 1624 struct f2fs_stat_info *stat_info; /* FS status information */ 1625 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1626 unsigned int segment_count[2]; /* # of allocated segments */ 1627 unsigned int block_count[2]; /* # of allocated blocks */ 1628 atomic_t inplace_count; /* # of inplace update */ 1629 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1630 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1631 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1632 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1633 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1634 atomic_t inline_inode; /* # of inline_data inodes */ 1635 atomic_t inline_dir; /* # of inline_dentry inodes */ 1636 atomic_t compr_inode; /* # of compressed inodes */ 1637 atomic64_t compr_blocks; /* # of compressed blocks */ 1638 atomic_t vw_cnt; /* # of volatile writes */ 1639 atomic_t max_aw_cnt; /* max # of atomic writes */ 1640 atomic_t max_vw_cnt; /* max # of volatile writes */ 1641 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1642 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1643 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1644 #endif 1645 spinlock_t stat_lock; /* lock for stat operations */ 1646 1647 /* For app/fs IO statistics */ 1648 spinlock_t iostat_lock; 1649 unsigned long long rw_iostat[NR_IO_TYPE]; 1650 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1651 bool iostat_enable; 1652 unsigned long iostat_next_period; 1653 unsigned int iostat_period_ms; 1654 1655 /* to attach REQ_META|REQ_FUA flags */ 1656 unsigned int data_io_flag; 1657 unsigned int node_io_flag; 1658 1659 /* For sysfs suppport */ 1660 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1661 struct completion s_kobj_unregister; 1662 1663 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1664 struct completion s_stat_kobj_unregister; 1665 1666 /* For shrinker support */ 1667 struct list_head s_list; 1668 int s_ndevs; /* number of devices */ 1669 struct f2fs_dev_info *devs; /* for device list */ 1670 unsigned int dirty_device; /* for checkpoint data flush */ 1671 spinlock_t dev_lock; /* protect dirty_device */ 1672 struct mutex umount_mutex; 1673 unsigned int shrinker_run_no; 1674 1675 /* For write statistics */ 1676 u64 sectors_written_start; 1677 u64 kbytes_written; 1678 1679 /* Reference to checksum algorithm driver via cryptoapi */ 1680 struct crypto_shash *s_chksum_driver; 1681 1682 /* Precomputed FS UUID checksum for seeding other checksums */ 1683 __u32 s_chksum_seed; 1684 1685 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1686 1687 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1688 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1689 1690 #ifdef CONFIG_F2FS_FS_COMPRESSION 1691 struct kmem_cache *page_array_slab; /* page array entry */ 1692 unsigned int page_array_slab_size; /* default page array slab size */ 1693 1694 /* For runtime compression statistics */ 1695 u64 compr_written_block; 1696 u64 compr_saved_block; 1697 u32 compr_new_inode; 1698 #endif 1699 }; 1700 1701 struct f2fs_private_dio { 1702 struct inode *inode; 1703 void *orig_private; 1704 bio_end_io_t *orig_end_io; 1705 bool write; 1706 }; 1707 1708 #ifdef CONFIG_F2FS_FAULT_INJECTION 1709 #define f2fs_show_injection_info(sbi, type) \ 1710 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1711 KERN_INFO, sbi->sb->s_id, \ 1712 f2fs_fault_name[type], \ 1713 __func__, __builtin_return_address(0)) 1714 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1715 { 1716 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1717 1718 if (!ffi->inject_rate) 1719 return false; 1720 1721 if (!IS_FAULT_SET(ffi, type)) 1722 return false; 1723 1724 atomic_inc(&ffi->inject_ops); 1725 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1726 atomic_set(&ffi->inject_ops, 0); 1727 return true; 1728 } 1729 return false; 1730 } 1731 #else 1732 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1733 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1734 { 1735 return false; 1736 } 1737 #endif 1738 1739 /* 1740 * Test if the mounted volume is a multi-device volume. 1741 * - For a single regular disk volume, sbi->s_ndevs is 0. 1742 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1743 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1744 */ 1745 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1746 { 1747 return sbi->s_ndevs > 1; 1748 } 1749 1750 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1751 { 1752 unsigned long now = jiffies; 1753 1754 sbi->last_time[type] = now; 1755 1756 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1757 if (type == REQ_TIME) { 1758 sbi->last_time[DISCARD_TIME] = now; 1759 sbi->last_time[GC_TIME] = now; 1760 } 1761 } 1762 1763 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1764 { 1765 unsigned long interval = sbi->interval_time[type] * HZ; 1766 1767 return time_after(jiffies, sbi->last_time[type] + interval); 1768 } 1769 1770 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1771 int type) 1772 { 1773 unsigned long interval = sbi->interval_time[type] * HZ; 1774 unsigned int wait_ms = 0; 1775 long delta; 1776 1777 delta = (sbi->last_time[type] + interval) - jiffies; 1778 if (delta > 0) 1779 wait_ms = jiffies_to_msecs(delta); 1780 1781 return wait_ms; 1782 } 1783 1784 /* 1785 * Inline functions 1786 */ 1787 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1788 const void *address, unsigned int length) 1789 { 1790 struct { 1791 struct shash_desc shash; 1792 char ctx[4]; 1793 } desc; 1794 int err; 1795 1796 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1797 1798 desc.shash.tfm = sbi->s_chksum_driver; 1799 *(u32 *)desc.ctx = crc; 1800 1801 err = crypto_shash_update(&desc.shash, address, length); 1802 BUG_ON(err); 1803 1804 return *(u32 *)desc.ctx; 1805 } 1806 1807 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1808 unsigned int length) 1809 { 1810 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1811 } 1812 1813 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1814 void *buf, size_t buf_size) 1815 { 1816 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1817 } 1818 1819 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1820 const void *address, unsigned int length) 1821 { 1822 return __f2fs_crc32(sbi, crc, address, length); 1823 } 1824 1825 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1826 { 1827 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1828 } 1829 1830 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1831 { 1832 return sb->s_fs_info; 1833 } 1834 1835 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1836 { 1837 return F2FS_SB(inode->i_sb); 1838 } 1839 1840 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1841 { 1842 return F2FS_I_SB(mapping->host); 1843 } 1844 1845 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1846 { 1847 return F2FS_M_SB(page_file_mapping(page)); 1848 } 1849 1850 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1851 { 1852 return (struct f2fs_super_block *)(sbi->raw_super); 1853 } 1854 1855 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1856 { 1857 return (struct f2fs_checkpoint *)(sbi->ckpt); 1858 } 1859 1860 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1861 { 1862 return (struct f2fs_node *)page_address(page); 1863 } 1864 1865 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1866 { 1867 return &((struct f2fs_node *)page_address(page))->i; 1868 } 1869 1870 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1871 { 1872 return (struct f2fs_nm_info *)(sbi->nm_info); 1873 } 1874 1875 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1876 { 1877 return (struct f2fs_sm_info *)(sbi->sm_info); 1878 } 1879 1880 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1881 { 1882 return (struct sit_info *)(SM_I(sbi)->sit_info); 1883 } 1884 1885 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1886 { 1887 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1888 } 1889 1890 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1891 { 1892 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1893 } 1894 1895 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1896 { 1897 return sbi->meta_inode->i_mapping; 1898 } 1899 1900 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1901 { 1902 return sbi->node_inode->i_mapping; 1903 } 1904 1905 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1906 { 1907 return test_bit(type, &sbi->s_flag); 1908 } 1909 1910 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1911 { 1912 set_bit(type, &sbi->s_flag); 1913 } 1914 1915 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1916 { 1917 clear_bit(type, &sbi->s_flag); 1918 } 1919 1920 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1921 { 1922 return le64_to_cpu(cp->checkpoint_ver); 1923 } 1924 1925 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1926 { 1927 if (type < F2FS_MAX_QUOTAS) 1928 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1929 return 0; 1930 } 1931 1932 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1933 { 1934 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1935 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1936 } 1937 1938 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1939 { 1940 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1941 1942 return ckpt_flags & f; 1943 } 1944 1945 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1946 { 1947 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1948 } 1949 1950 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1951 { 1952 unsigned int ckpt_flags; 1953 1954 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1955 ckpt_flags |= f; 1956 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1957 } 1958 1959 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1960 { 1961 unsigned long flags; 1962 1963 spin_lock_irqsave(&sbi->cp_lock, flags); 1964 __set_ckpt_flags(F2FS_CKPT(sbi), f); 1965 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1966 } 1967 1968 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1969 { 1970 unsigned int ckpt_flags; 1971 1972 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1973 ckpt_flags &= (~f); 1974 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1975 } 1976 1977 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1978 { 1979 unsigned long flags; 1980 1981 spin_lock_irqsave(&sbi->cp_lock, flags); 1982 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 1983 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1984 } 1985 1986 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 1987 { 1988 unsigned long flags; 1989 unsigned char *nat_bits; 1990 1991 /* 1992 * In order to re-enable nat_bits we need to call fsck.f2fs by 1993 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 1994 * so let's rely on regular fsck or unclean shutdown. 1995 */ 1996 1997 if (lock) 1998 spin_lock_irqsave(&sbi->cp_lock, flags); 1999 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 2000 nat_bits = NM_I(sbi)->nat_bits; 2001 NM_I(sbi)->nat_bits = NULL; 2002 if (lock) 2003 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2004 2005 kvfree(nat_bits); 2006 } 2007 2008 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 2009 struct cp_control *cpc) 2010 { 2011 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 2012 2013 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 2014 } 2015 2016 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2017 { 2018 down_read(&sbi->cp_rwsem); 2019 } 2020 2021 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2022 { 2023 return down_read_trylock(&sbi->cp_rwsem); 2024 } 2025 2026 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2027 { 2028 up_read(&sbi->cp_rwsem); 2029 } 2030 2031 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2032 { 2033 down_write(&sbi->cp_rwsem); 2034 } 2035 2036 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2037 { 2038 up_write(&sbi->cp_rwsem); 2039 } 2040 2041 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2042 { 2043 int reason = CP_SYNC; 2044 2045 if (test_opt(sbi, FASTBOOT)) 2046 reason = CP_FASTBOOT; 2047 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2048 reason = CP_UMOUNT; 2049 return reason; 2050 } 2051 2052 static inline bool __remain_node_summaries(int reason) 2053 { 2054 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2055 } 2056 2057 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2058 { 2059 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2060 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2061 } 2062 2063 /* 2064 * Check whether the inode has blocks or not 2065 */ 2066 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2067 { 2068 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2069 2070 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2071 } 2072 2073 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2074 { 2075 return ofs == XATTR_NODE_OFFSET; 2076 } 2077 2078 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2079 struct inode *inode, bool cap) 2080 { 2081 if (!inode) 2082 return true; 2083 if (!test_opt(sbi, RESERVE_ROOT)) 2084 return false; 2085 if (IS_NOQUOTA(inode)) 2086 return true; 2087 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2088 return true; 2089 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2090 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2091 return true; 2092 if (cap && capable(CAP_SYS_RESOURCE)) 2093 return true; 2094 return false; 2095 } 2096 2097 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2098 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2099 struct inode *inode, blkcnt_t *count) 2100 { 2101 blkcnt_t diff = 0, release = 0; 2102 block_t avail_user_block_count; 2103 int ret; 2104 2105 ret = dquot_reserve_block(inode, *count); 2106 if (ret) 2107 return ret; 2108 2109 if (time_to_inject(sbi, FAULT_BLOCK)) { 2110 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2111 release = *count; 2112 goto release_quota; 2113 } 2114 2115 /* 2116 * let's increase this in prior to actual block count change in order 2117 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2118 */ 2119 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2120 2121 spin_lock(&sbi->stat_lock); 2122 sbi->total_valid_block_count += (block_t)(*count); 2123 avail_user_block_count = sbi->user_block_count - 2124 sbi->current_reserved_blocks; 2125 2126 if (!__allow_reserved_blocks(sbi, inode, true)) 2127 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2128 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2129 if (avail_user_block_count > sbi->unusable_block_count) 2130 avail_user_block_count -= sbi->unusable_block_count; 2131 else 2132 avail_user_block_count = 0; 2133 } 2134 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2135 diff = sbi->total_valid_block_count - avail_user_block_count; 2136 if (diff > *count) 2137 diff = *count; 2138 *count -= diff; 2139 release = diff; 2140 sbi->total_valid_block_count -= diff; 2141 if (!*count) { 2142 spin_unlock(&sbi->stat_lock); 2143 goto enospc; 2144 } 2145 } 2146 spin_unlock(&sbi->stat_lock); 2147 2148 if (unlikely(release)) { 2149 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2150 dquot_release_reservation_block(inode, release); 2151 } 2152 f2fs_i_blocks_write(inode, *count, true, true); 2153 return 0; 2154 2155 enospc: 2156 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2157 release_quota: 2158 dquot_release_reservation_block(inode, release); 2159 return -ENOSPC; 2160 } 2161 2162 __printf(2, 3) 2163 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2164 2165 #define f2fs_err(sbi, fmt, ...) \ 2166 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2167 #define f2fs_warn(sbi, fmt, ...) \ 2168 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2169 #define f2fs_notice(sbi, fmt, ...) \ 2170 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2171 #define f2fs_info(sbi, fmt, ...) \ 2172 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2173 #define f2fs_debug(sbi, fmt, ...) \ 2174 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2175 2176 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2177 struct inode *inode, 2178 block_t count) 2179 { 2180 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2181 2182 spin_lock(&sbi->stat_lock); 2183 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2184 sbi->total_valid_block_count -= (block_t)count; 2185 if (sbi->reserved_blocks && 2186 sbi->current_reserved_blocks < sbi->reserved_blocks) 2187 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2188 sbi->current_reserved_blocks + count); 2189 spin_unlock(&sbi->stat_lock); 2190 if (unlikely(inode->i_blocks < sectors)) { 2191 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2192 inode->i_ino, 2193 (unsigned long long)inode->i_blocks, 2194 (unsigned long long)sectors); 2195 set_sbi_flag(sbi, SBI_NEED_FSCK); 2196 return; 2197 } 2198 f2fs_i_blocks_write(inode, count, false, true); 2199 } 2200 2201 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2202 { 2203 atomic_inc(&sbi->nr_pages[count_type]); 2204 2205 if (count_type == F2FS_DIRTY_DENTS || 2206 count_type == F2FS_DIRTY_NODES || 2207 count_type == F2FS_DIRTY_META || 2208 count_type == F2FS_DIRTY_QDATA || 2209 count_type == F2FS_DIRTY_IMETA) 2210 set_sbi_flag(sbi, SBI_IS_DIRTY); 2211 } 2212 2213 static inline void inode_inc_dirty_pages(struct inode *inode) 2214 { 2215 atomic_inc(&F2FS_I(inode)->dirty_pages); 2216 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2217 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2218 if (IS_NOQUOTA(inode)) 2219 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2220 } 2221 2222 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2223 { 2224 atomic_dec(&sbi->nr_pages[count_type]); 2225 } 2226 2227 static inline void inode_dec_dirty_pages(struct inode *inode) 2228 { 2229 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2230 !S_ISLNK(inode->i_mode)) 2231 return; 2232 2233 atomic_dec(&F2FS_I(inode)->dirty_pages); 2234 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2235 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2236 if (IS_NOQUOTA(inode)) 2237 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2238 } 2239 2240 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2241 { 2242 return atomic_read(&sbi->nr_pages[count_type]); 2243 } 2244 2245 static inline int get_dirty_pages(struct inode *inode) 2246 { 2247 return atomic_read(&F2FS_I(inode)->dirty_pages); 2248 } 2249 2250 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2251 { 2252 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2253 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2254 sbi->log_blocks_per_seg; 2255 2256 return segs / sbi->segs_per_sec; 2257 } 2258 2259 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2260 { 2261 return sbi->total_valid_block_count; 2262 } 2263 2264 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2265 { 2266 return sbi->discard_blks; 2267 } 2268 2269 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2270 { 2271 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2272 2273 /* return NAT or SIT bitmap */ 2274 if (flag == NAT_BITMAP) 2275 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2276 else if (flag == SIT_BITMAP) 2277 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2278 2279 return 0; 2280 } 2281 2282 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2283 { 2284 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2285 } 2286 2287 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2288 { 2289 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2290 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2291 int offset; 2292 2293 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2294 offset = (flag == SIT_BITMAP) ? 2295 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2296 /* 2297 * if large_nat_bitmap feature is enabled, leave checksum 2298 * protection for all nat/sit bitmaps. 2299 */ 2300 return tmp_ptr + offset + sizeof(__le32); 2301 } 2302 2303 if (__cp_payload(sbi) > 0) { 2304 if (flag == NAT_BITMAP) 2305 return &ckpt->sit_nat_version_bitmap; 2306 else 2307 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2308 } else { 2309 offset = (flag == NAT_BITMAP) ? 2310 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2311 return tmp_ptr + offset; 2312 } 2313 } 2314 2315 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2316 { 2317 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2318 2319 if (sbi->cur_cp_pack == 2) 2320 start_addr += sbi->blocks_per_seg; 2321 return start_addr; 2322 } 2323 2324 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2325 { 2326 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2327 2328 if (sbi->cur_cp_pack == 1) 2329 start_addr += sbi->blocks_per_seg; 2330 return start_addr; 2331 } 2332 2333 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2334 { 2335 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2336 } 2337 2338 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2339 { 2340 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2341 } 2342 2343 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2344 struct inode *inode, bool is_inode) 2345 { 2346 block_t valid_block_count; 2347 unsigned int valid_node_count, user_block_count; 2348 int err; 2349 2350 if (is_inode) { 2351 if (inode) { 2352 err = dquot_alloc_inode(inode); 2353 if (err) 2354 return err; 2355 } 2356 } else { 2357 err = dquot_reserve_block(inode, 1); 2358 if (err) 2359 return err; 2360 } 2361 2362 if (time_to_inject(sbi, FAULT_BLOCK)) { 2363 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2364 goto enospc; 2365 } 2366 2367 spin_lock(&sbi->stat_lock); 2368 2369 valid_block_count = sbi->total_valid_block_count + 2370 sbi->current_reserved_blocks + 1; 2371 2372 if (!__allow_reserved_blocks(sbi, inode, false)) 2373 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2374 user_block_count = sbi->user_block_count; 2375 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2376 user_block_count -= sbi->unusable_block_count; 2377 2378 if (unlikely(valid_block_count > user_block_count)) { 2379 spin_unlock(&sbi->stat_lock); 2380 goto enospc; 2381 } 2382 2383 valid_node_count = sbi->total_valid_node_count + 1; 2384 if (unlikely(valid_node_count > sbi->total_node_count)) { 2385 spin_unlock(&sbi->stat_lock); 2386 goto enospc; 2387 } 2388 2389 sbi->total_valid_node_count++; 2390 sbi->total_valid_block_count++; 2391 spin_unlock(&sbi->stat_lock); 2392 2393 if (inode) { 2394 if (is_inode) 2395 f2fs_mark_inode_dirty_sync(inode, true); 2396 else 2397 f2fs_i_blocks_write(inode, 1, true, true); 2398 } 2399 2400 percpu_counter_inc(&sbi->alloc_valid_block_count); 2401 return 0; 2402 2403 enospc: 2404 if (is_inode) { 2405 if (inode) 2406 dquot_free_inode(inode); 2407 } else { 2408 dquot_release_reservation_block(inode, 1); 2409 } 2410 return -ENOSPC; 2411 } 2412 2413 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2414 struct inode *inode, bool is_inode) 2415 { 2416 spin_lock(&sbi->stat_lock); 2417 2418 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2419 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2420 2421 sbi->total_valid_node_count--; 2422 sbi->total_valid_block_count--; 2423 if (sbi->reserved_blocks && 2424 sbi->current_reserved_blocks < sbi->reserved_blocks) 2425 sbi->current_reserved_blocks++; 2426 2427 spin_unlock(&sbi->stat_lock); 2428 2429 if (is_inode) { 2430 dquot_free_inode(inode); 2431 } else { 2432 if (unlikely(inode->i_blocks == 0)) { 2433 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2434 inode->i_ino, 2435 (unsigned long long)inode->i_blocks); 2436 set_sbi_flag(sbi, SBI_NEED_FSCK); 2437 return; 2438 } 2439 f2fs_i_blocks_write(inode, 1, false, true); 2440 } 2441 } 2442 2443 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2444 { 2445 return sbi->total_valid_node_count; 2446 } 2447 2448 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2449 { 2450 percpu_counter_inc(&sbi->total_valid_inode_count); 2451 } 2452 2453 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2454 { 2455 percpu_counter_dec(&sbi->total_valid_inode_count); 2456 } 2457 2458 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2459 { 2460 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2461 } 2462 2463 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2464 pgoff_t index, bool for_write) 2465 { 2466 struct page *page; 2467 2468 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2469 if (!for_write) 2470 page = find_get_page_flags(mapping, index, 2471 FGP_LOCK | FGP_ACCESSED); 2472 else 2473 page = find_lock_page(mapping, index); 2474 if (page) 2475 return page; 2476 2477 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2478 f2fs_show_injection_info(F2FS_M_SB(mapping), 2479 FAULT_PAGE_ALLOC); 2480 return NULL; 2481 } 2482 } 2483 2484 if (!for_write) 2485 return grab_cache_page(mapping, index); 2486 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2487 } 2488 2489 static inline struct page *f2fs_pagecache_get_page( 2490 struct address_space *mapping, pgoff_t index, 2491 int fgp_flags, gfp_t gfp_mask) 2492 { 2493 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2494 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2495 return NULL; 2496 } 2497 2498 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2499 } 2500 2501 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2502 { 2503 char *src_kaddr = kmap(src); 2504 char *dst_kaddr = kmap(dst); 2505 2506 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2507 kunmap(dst); 2508 kunmap(src); 2509 } 2510 2511 static inline void f2fs_put_page(struct page *page, int unlock) 2512 { 2513 if (!page) 2514 return; 2515 2516 if (unlock) { 2517 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2518 unlock_page(page); 2519 } 2520 put_page(page); 2521 } 2522 2523 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2524 { 2525 if (dn->node_page) 2526 f2fs_put_page(dn->node_page, 1); 2527 if (dn->inode_page && dn->node_page != dn->inode_page) 2528 f2fs_put_page(dn->inode_page, 0); 2529 dn->node_page = NULL; 2530 dn->inode_page = NULL; 2531 } 2532 2533 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2534 size_t size) 2535 { 2536 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2537 } 2538 2539 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2540 gfp_t flags) 2541 { 2542 void *entry; 2543 2544 entry = kmem_cache_alloc(cachep, flags); 2545 if (!entry) 2546 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2547 return entry; 2548 } 2549 2550 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2551 { 2552 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2553 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2554 get_pages(sbi, F2FS_WB_CP_DATA) || 2555 get_pages(sbi, F2FS_DIO_READ) || 2556 get_pages(sbi, F2FS_DIO_WRITE)) 2557 return true; 2558 2559 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2560 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2561 return true; 2562 2563 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2564 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2565 return true; 2566 return false; 2567 } 2568 2569 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2570 { 2571 if (sbi->gc_mode == GC_URGENT_HIGH) 2572 return true; 2573 2574 if (is_inflight_io(sbi, type)) 2575 return false; 2576 2577 if (sbi->gc_mode == GC_URGENT_LOW && 2578 (type == DISCARD_TIME || type == GC_TIME)) 2579 return true; 2580 2581 return f2fs_time_over(sbi, type); 2582 } 2583 2584 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2585 unsigned long index, void *item) 2586 { 2587 while (radix_tree_insert(root, index, item)) 2588 cond_resched(); 2589 } 2590 2591 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2592 2593 static inline bool IS_INODE(struct page *page) 2594 { 2595 struct f2fs_node *p = F2FS_NODE(page); 2596 2597 return RAW_IS_INODE(p); 2598 } 2599 2600 static inline int offset_in_addr(struct f2fs_inode *i) 2601 { 2602 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2603 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2604 } 2605 2606 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2607 { 2608 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2609 } 2610 2611 static inline int f2fs_has_extra_attr(struct inode *inode); 2612 static inline block_t data_blkaddr(struct inode *inode, 2613 struct page *node_page, unsigned int offset) 2614 { 2615 struct f2fs_node *raw_node; 2616 __le32 *addr_array; 2617 int base = 0; 2618 bool is_inode = IS_INODE(node_page); 2619 2620 raw_node = F2FS_NODE(node_page); 2621 2622 if (is_inode) { 2623 if (!inode) 2624 /* from GC path only */ 2625 base = offset_in_addr(&raw_node->i); 2626 else if (f2fs_has_extra_attr(inode)) 2627 base = get_extra_isize(inode); 2628 } 2629 2630 addr_array = blkaddr_in_node(raw_node); 2631 return le32_to_cpu(addr_array[base + offset]); 2632 } 2633 2634 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2635 { 2636 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2637 } 2638 2639 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2640 { 2641 int mask; 2642 2643 addr += (nr >> 3); 2644 mask = 1 << (7 - (nr & 0x07)); 2645 return mask & *addr; 2646 } 2647 2648 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2649 { 2650 int mask; 2651 2652 addr += (nr >> 3); 2653 mask = 1 << (7 - (nr & 0x07)); 2654 *addr |= mask; 2655 } 2656 2657 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2658 { 2659 int mask; 2660 2661 addr += (nr >> 3); 2662 mask = 1 << (7 - (nr & 0x07)); 2663 *addr &= ~mask; 2664 } 2665 2666 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2667 { 2668 int mask; 2669 int ret; 2670 2671 addr += (nr >> 3); 2672 mask = 1 << (7 - (nr & 0x07)); 2673 ret = mask & *addr; 2674 *addr |= mask; 2675 return ret; 2676 } 2677 2678 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2679 { 2680 int mask; 2681 int ret; 2682 2683 addr += (nr >> 3); 2684 mask = 1 << (7 - (nr & 0x07)); 2685 ret = mask & *addr; 2686 *addr &= ~mask; 2687 return ret; 2688 } 2689 2690 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2691 { 2692 int mask; 2693 2694 addr += (nr >> 3); 2695 mask = 1 << (7 - (nr & 0x07)); 2696 *addr ^= mask; 2697 } 2698 2699 /* 2700 * On-disk inode flags (f2fs_inode::i_flags) 2701 */ 2702 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2703 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2704 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2705 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2706 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2707 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2708 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2709 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2710 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2711 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2712 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2713 2714 /* Flags that should be inherited by new inodes from their parent. */ 2715 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2716 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2717 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2718 2719 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2720 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2721 F2FS_CASEFOLD_FL)) 2722 2723 /* Flags that are appropriate for non-directories/regular files. */ 2724 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2725 2726 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2727 { 2728 if (S_ISDIR(mode)) 2729 return flags; 2730 else if (S_ISREG(mode)) 2731 return flags & F2FS_REG_FLMASK; 2732 else 2733 return flags & F2FS_OTHER_FLMASK; 2734 } 2735 2736 static inline void __mark_inode_dirty_flag(struct inode *inode, 2737 int flag, bool set) 2738 { 2739 switch (flag) { 2740 case FI_INLINE_XATTR: 2741 case FI_INLINE_DATA: 2742 case FI_INLINE_DENTRY: 2743 case FI_NEW_INODE: 2744 if (set) 2745 return; 2746 fallthrough; 2747 case FI_DATA_EXIST: 2748 case FI_INLINE_DOTS: 2749 case FI_PIN_FILE: 2750 case FI_COMPRESS_RELEASED: 2751 f2fs_mark_inode_dirty_sync(inode, true); 2752 } 2753 } 2754 2755 static inline void set_inode_flag(struct inode *inode, int flag) 2756 { 2757 set_bit(flag, F2FS_I(inode)->flags); 2758 __mark_inode_dirty_flag(inode, flag, true); 2759 } 2760 2761 static inline int is_inode_flag_set(struct inode *inode, int flag) 2762 { 2763 return test_bit(flag, F2FS_I(inode)->flags); 2764 } 2765 2766 static inline void clear_inode_flag(struct inode *inode, int flag) 2767 { 2768 clear_bit(flag, F2FS_I(inode)->flags); 2769 __mark_inode_dirty_flag(inode, flag, false); 2770 } 2771 2772 static inline bool f2fs_verity_in_progress(struct inode *inode) 2773 { 2774 return IS_ENABLED(CONFIG_FS_VERITY) && 2775 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2776 } 2777 2778 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2779 { 2780 F2FS_I(inode)->i_acl_mode = mode; 2781 set_inode_flag(inode, FI_ACL_MODE); 2782 f2fs_mark_inode_dirty_sync(inode, false); 2783 } 2784 2785 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2786 { 2787 if (inc) 2788 inc_nlink(inode); 2789 else 2790 drop_nlink(inode); 2791 f2fs_mark_inode_dirty_sync(inode, true); 2792 } 2793 2794 static inline void f2fs_i_blocks_write(struct inode *inode, 2795 block_t diff, bool add, bool claim) 2796 { 2797 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2798 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2799 2800 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2801 if (add) { 2802 if (claim) 2803 dquot_claim_block(inode, diff); 2804 else 2805 dquot_alloc_block_nofail(inode, diff); 2806 } else { 2807 dquot_free_block(inode, diff); 2808 } 2809 2810 f2fs_mark_inode_dirty_sync(inode, true); 2811 if (clean || recover) 2812 set_inode_flag(inode, FI_AUTO_RECOVER); 2813 } 2814 2815 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2816 { 2817 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2818 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2819 2820 if (i_size_read(inode) == i_size) 2821 return; 2822 2823 i_size_write(inode, i_size); 2824 f2fs_mark_inode_dirty_sync(inode, true); 2825 if (clean || recover) 2826 set_inode_flag(inode, FI_AUTO_RECOVER); 2827 } 2828 2829 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2830 { 2831 F2FS_I(inode)->i_current_depth = depth; 2832 f2fs_mark_inode_dirty_sync(inode, true); 2833 } 2834 2835 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2836 unsigned int count) 2837 { 2838 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2839 f2fs_mark_inode_dirty_sync(inode, true); 2840 } 2841 2842 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2843 { 2844 F2FS_I(inode)->i_xattr_nid = xnid; 2845 f2fs_mark_inode_dirty_sync(inode, true); 2846 } 2847 2848 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2849 { 2850 F2FS_I(inode)->i_pino = pino; 2851 f2fs_mark_inode_dirty_sync(inode, true); 2852 } 2853 2854 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2855 { 2856 struct f2fs_inode_info *fi = F2FS_I(inode); 2857 2858 if (ri->i_inline & F2FS_INLINE_XATTR) 2859 set_bit(FI_INLINE_XATTR, fi->flags); 2860 if (ri->i_inline & F2FS_INLINE_DATA) 2861 set_bit(FI_INLINE_DATA, fi->flags); 2862 if (ri->i_inline & F2FS_INLINE_DENTRY) 2863 set_bit(FI_INLINE_DENTRY, fi->flags); 2864 if (ri->i_inline & F2FS_DATA_EXIST) 2865 set_bit(FI_DATA_EXIST, fi->flags); 2866 if (ri->i_inline & F2FS_INLINE_DOTS) 2867 set_bit(FI_INLINE_DOTS, fi->flags); 2868 if (ri->i_inline & F2FS_EXTRA_ATTR) 2869 set_bit(FI_EXTRA_ATTR, fi->flags); 2870 if (ri->i_inline & F2FS_PIN_FILE) 2871 set_bit(FI_PIN_FILE, fi->flags); 2872 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 2873 set_bit(FI_COMPRESS_RELEASED, fi->flags); 2874 } 2875 2876 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2877 { 2878 ri->i_inline = 0; 2879 2880 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2881 ri->i_inline |= F2FS_INLINE_XATTR; 2882 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2883 ri->i_inline |= F2FS_INLINE_DATA; 2884 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2885 ri->i_inline |= F2FS_INLINE_DENTRY; 2886 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2887 ri->i_inline |= F2FS_DATA_EXIST; 2888 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2889 ri->i_inline |= F2FS_INLINE_DOTS; 2890 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2891 ri->i_inline |= F2FS_EXTRA_ATTR; 2892 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2893 ri->i_inline |= F2FS_PIN_FILE; 2894 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 2895 ri->i_inline |= F2FS_COMPRESS_RELEASED; 2896 } 2897 2898 static inline int f2fs_has_extra_attr(struct inode *inode) 2899 { 2900 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2901 } 2902 2903 static inline int f2fs_has_inline_xattr(struct inode *inode) 2904 { 2905 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2906 } 2907 2908 static inline int f2fs_compressed_file(struct inode *inode) 2909 { 2910 return S_ISREG(inode->i_mode) && 2911 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2912 } 2913 2914 static inline bool f2fs_need_compress_data(struct inode *inode) 2915 { 2916 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 2917 2918 if (!f2fs_compressed_file(inode)) 2919 return false; 2920 2921 if (compress_mode == COMPR_MODE_FS) 2922 return true; 2923 else if (compress_mode == COMPR_MODE_USER && 2924 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 2925 return true; 2926 2927 return false; 2928 } 2929 2930 static inline unsigned int addrs_per_inode(struct inode *inode) 2931 { 2932 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2933 get_inline_xattr_addrs(inode); 2934 2935 if (!f2fs_compressed_file(inode)) 2936 return addrs; 2937 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 2938 } 2939 2940 static inline unsigned int addrs_per_block(struct inode *inode) 2941 { 2942 if (!f2fs_compressed_file(inode)) 2943 return DEF_ADDRS_PER_BLOCK; 2944 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 2945 } 2946 2947 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2948 { 2949 struct f2fs_inode *ri = F2FS_INODE(page); 2950 2951 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2952 get_inline_xattr_addrs(inode)]); 2953 } 2954 2955 static inline int inline_xattr_size(struct inode *inode) 2956 { 2957 if (f2fs_has_inline_xattr(inode)) 2958 return get_inline_xattr_addrs(inode) * sizeof(__le32); 2959 return 0; 2960 } 2961 2962 static inline int f2fs_has_inline_data(struct inode *inode) 2963 { 2964 return is_inode_flag_set(inode, FI_INLINE_DATA); 2965 } 2966 2967 static inline int f2fs_exist_data(struct inode *inode) 2968 { 2969 return is_inode_flag_set(inode, FI_DATA_EXIST); 2970 } 2971 2972 static inline int f2fs_has_inline_dots(struct inode *inode) 2973 { 2974 return is_inode_flag_set(inode, FI_INLINE_DOTS); 2975 } 2976 2977 static inline int f2fs_is_mmap_file(struct inode *inode) 2978 { 2979 return is_inode_flag_set(inode, FI_MMAP_FILE); 2980 } 2981 2982 static inline bool f2fs_is_pinned_file(struct inode *inode) 2983 { 2984 return is_inode_flag_set(inode, FI_PIN_FILE); 2985 } 2986 2987 static inline bool f2fs_is_atomic_file(struct inode *inode) 2988 { 2989 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 2990 } 2991 2992 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 2993 { 2994 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 2995 } 2996 2997 static inline bool f2fs_is_volatile_file(struct inode *inode) 2998 { 2999 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 3000 } 3001 3002 static inline bool f2fs_is_first_block_written(struct inode *inode) 3003 { 3004 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3005 } 3006 3007 static inline bool f2fs_is_drop_cache(struct inode *inode) 3008 { 3009 return is_inode_flag_set(inode, FI_DROP_CACHE); 3010 } 3011 3012 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3013 { 3014 struct f2fs_inode *ri = F2FS_INODE(page); 3015 int extra_size = get_extra_isize(inode); 3016 3017 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3018 } 3019 3020 static inline int f2fs_has_inline_dentry(struct inode *inode) 3021 { 3022 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3023 } 3024 3025 static inline int is_file(struct inode *inode, int type) 3026 { 3027 return F2FS_I(inode)->i_advise & type; 3028 } 3029 3030 static inline void set_file(struct inode *inode, int type) 3031 { 3032 F2FS_I(inode)->i_advise |= type; 3033 f2fs_mark_inode_dirty_sync(inode, true); 3034 } 3035 3036 static inline void clear_file(struct inode *inode, int type) 3037 { 3038 F2FS_I(inode)->i_advise &= ~type; 3039 f2fs_mark_inode_dirty_sync(inode, true); 3040 } 3041 3042 static inline bool f2fs_is_time_consistent(struct inode *inode) 3043 { 3044 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3045 return false; 3046 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3047 return false; 3048 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3049 return false; 3050 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3051 &F2FS_I(inode)->i_crtime)) 3052 return false; 3053 return true; 3054 } 3055 3056 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3057 { 3058 bool ret; 3059 3060 if (dsync) { 3061 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3062 3063 spin_lock(&sbi->inode_lock[DIRTY_META]); 3064 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3065 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3066 return ret; 3067 } 3068 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3069 file_keep_isize(inode) || 3070 i_size_read(inode) & ~PAGE_MASK) 3071 return false; 3072 3073 if (!f2fs_is_time_consistent(inode)) 3074 return false; 3075 3076 spin_lock(&F2FS_I(inode)->i_size_lock); 3077 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3078 spin_unlock(&F2FS_I(inode)->i_size_lock); 3079 3080 return ret; 3081 } 3082 3083 static inline bool f2fs_readonly(struct super_block *sb) 3084 { 3085 return sb_rdonly(sb); 3086 } 3087 3088 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3089 { 3090 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3091 } 3092 3093 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3094 { 3095 if (len == 1 && name[0] == '.') 3096 return true; 3097 3098 if (len == 2 && name[0] == '.' && name[1] == '.') 3099 return true; 3100 3101 return false; 3102 } 3103 3104 static inline bool f2fs_may_extent_tree(struct inode *inode) 3105 { 3106 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3107 3108 if (!test_opt(sbi, EXTENT_CACHE) || 3109 is_inode_flag_set(inode, FI_NO_EXTENT) || 3110 is_inode_flag_set(inode, FI_COMPRESSED_FILE)) 3111 return false; 3112 3113 /* 3114 * for recovered files during mount do not create extents 3115 * if shrinker is not registered. 3116 */ 3117 if (list_empty(&sbi->s_list)) 3118 return false; 3119 3120 return S_ISREG(inode->i_mode); 3121 } 3122 3123 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3124 size_t size, gfp_t flags) 3125 { 3126 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3127 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3128 return NULL; 3129 } 3130 3131 return kmalloc(size, flags); 3132 } 3133 3134 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3135 size_t size, gfp_t flags) 3136 { 3137 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3138 } 3139 3140 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3141 size_t size, gfp_t flags) 3142 { 3143 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3144 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3145 return NULL; 3146 } 3147 3148 return kvmalloc(size, flags); 3149 } 3150 3151 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3152 size_t size, gfp_t flags) 3153 { 3154 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3155 } 3156 3157 static inline int get_extra_isize(struct inode *inode) 3158 { 3159 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3160 } 3161 3162 static inline int get_inline_xattr_addrs(struct inode *inode) 3163 { 3164 return F2FS_I(inode)->i_inline_xattr_size; 3165 } 3166 3167 #define f2fs_get_inode_mode(i) \ 3168 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3169 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3170 3171 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3172 (offsetof(struct f2fs_inode, i_extra_end) - \ 3173 offsetof(struct f2fs_inode, i_extra_isize)) \ 3174 3175 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3176 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3177 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3178 sizeof((f2fs_inode)->field)) \ 3179 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3180 3181 #define DEFAULT_IOSTAT_PERIOD_MS 3000 3182 #define MIN_IOSTAT_PERIOD_MS 100 3183 /* maximum period of iostat tracing is 1 day */ 3184 #define MAX_IOSTAT_PERIOD_MS 8640000 3185 3186 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 3187 { 3188 int i; 3189 3190 spin_lock(&sbi->iostat_lock); 3191 for (i = 0; i < NR_IO_TYPE; i++) { 3192 sbi->rw_iostat[i] = 0; 3193 sbi->prev_rw_iostat[i] = 0; 3194 } 3195 spin_unlock(&sbi->iostat_lock); 3196 } 3197 3198 extern void f2fs_record_iostat(struct f2fs_sb_info *sbi); 3199 3200 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 3201 enum iostat_type type, unsigned long long io_bytes) 3202 { 3203 if (!sbi->iostat_enable) 3204 return; 3205 spin_lock(&sbi->iostat_lock); 3206 sbi->rw_iostat[type] += io_bytes; 3207 3208 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 3209 sbi->rw_iostat[APP_BUFFERED_IO] = 3210 sbi->rw_iostat[APP_WRITE_IO] - 3211 sbi->rw_iostat[APP_DIRECT_IO]; 3212 3213 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO) 3214 sbi->rw_iostat[APP_BUFFERED_READ_IO] = 3215 sbi->rw_iostat[APP_READ_IO] - 3216 sbi->rw_iostat[APP_DIRECT_READ_IO]; 3217 spin_unlock(&sbi->iostat_lock); 3218 3219 f2fs_record_iostat(sbi); 3220 } 3221 3222 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3223 3224 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3225 3226 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3227 block_t blkaddr, int type); 3228 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3229 block_t blkaddr, int type) 3230 { 3231 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3232 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3233 blkaddr, type); 3234 f2fs_bug_on(sbi, 1); 3235 } 3236 } 3237 3238 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3239 { 3240 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3241 blkaddr == COMPRESS_ADDR) 3242 return false; 3243 return true; 3244 } 3245 3246 /* 3247 * file.c 3248 */ 3249 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3250 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3251 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3252 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3253 int f2fs_truncate(struct inode *inode); 3254 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3255 struct kstat *stat, u32 request_mask, unsigned int flags); 3256 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3257 struct iattr *attr); 3258 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3259 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3260 int f2fs_precache_extents(struct inode *inode); 3261 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3262 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3263 struct dentry *dentry, struct fileattr *fa); 3264 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3265 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3266 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3267 int f2fs_pin_file_control(struct inode *inode, bool inc); 3268 3269 /* 3270 * inode.c 3271 */ 3272 void f2fs_set_inode_flags(struct inode *inode); 3273 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3274 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3275 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3276 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3277 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3278 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3279 void f2fs_update_inode_page(struct inode *inode); 3280 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3281 void f2fs_evict_inode(struct inode *inode); 3282 void f2fs_handle_failed_inode(struct inode *inode); 3283 3284 /* 3285 * namei.c 3286 */ 3287 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3288 bool hot, bool set); 3289 struct dentry *f2fs_get_parent(struct dentry *child); 3290 3291 /* 3292 * dir.c 3293 */ 3294 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3295 int f2fs_init_casefolded_name(const struct inode *dir, 3296 struct f2fs_filename *fname); 3297 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3298 int lookup, struct f2fs_filename *fname); 3299 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3300 struct f2fs_filename *fname); 3301 void f2fs_free_filename(struct f2fs_filename *fname); 3302 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3303 const struct f2fs_filename *fname, int *max_slots); 3304 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3305 unsigned int start_pos, struct fscrypt_str *fstr); 3306 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3307 struct f2fs_dentry_ptr *d); 3308 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3309 const struct f2fs_filename *fname, struct page *dpage); 3310 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3311 unsigned int current_depth); 3312 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3313 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3314 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3315 const struct f2fs_filename *fname, 3316 struct page **res_page); 3317 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3318 const struct qstr *child, struct page **res_page); 3319 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3320 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3321 struct page **page); 3322 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3323 struct page *page, struct inode *inode); 3324 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3325 const struct f2fs_filename *fname); 3326 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3327 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3328 unsigned int bit_pos); 3329 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3330 struct inode *inode, nid_t ino, umode_t mode); 3331 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3332 struct inode *inode, nid_t ino, umode_t mode); 3333 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3334 struct inode *inode, nid_t ino, umode_t mode); 3335 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3336 struct inode *dir, struct inode *inode); 3337 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3338 bool f2fs_empty_dir(struct inode *dir); 3339 3340 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3341 { 3342 if (fscrypt_is_nokey_name(dentry)) 3343 return -ENOKEY; 3344 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3345 inode, inode->i_ino, inode->i_mode); 3346 } 3347 3348 /* 3349 * super.c 3350 */ 3351 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3352 void f2fs_inode_synced(struct inode *inode); 3353 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3354 int f2fs_quota_sync(struct super_block *sb, int type); 3355 loff_t max_file_blocks(struct inode *inode); 3356 void f2fs_quota_off_umount(struct super_block *sb); 3357 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3358 int f2fs_sync_fs(struct super_block *sb, int sync); 3359 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3360 3361 /* 3362 * hash.c 3363 */ 3364 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3365 3366 /* 3367 * node.c 3368 */ 3369 struct node_info; 3370 3371 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3372 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3373 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3374 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3375 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3376 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3377 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3378 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3379 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3380 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3381 struct node_info *ni); 3382 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3383 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3384 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3385 int f2fs_truncate_xattr_node(struct inode *inode); 3386 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3387 unsigned int seq_id); 3388 int f2fs_remove_inode_page(struct inode *inode); 3389 struct page *f2fs_new_inode_page(struct inode *inode); 3390 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3391 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3392 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3393 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3394 int f2fs_move_node_page(struct page *node_page, int gc_type); 3395 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3396 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3397 struct writeback_control *wbc, bool atomic, 3398 unsigned int *seq_id); 3399 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3400 struct writeback_control *wbc, 3401 bool do_balance, enum iostat_type io_type); 3402 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3403 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3404 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3405 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3406 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3407 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3408 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3409 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3410 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3411 unsigned int segno, struct f2fs_summary_block *sum); 3412 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3413 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3414 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3415 int __init f2fs_create_node_manager_caches(void); 3416 void f2fs_destroy_node_manager_caches(void); 3417 3418 /* 3419 * segment.c 3420 */ 3421 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3422 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3423 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3424 void f2fs_drop_inmem_pages(struct inode *inode); 3425 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3426 int f2fs_commit_inmem_pages(struct inode *inode); 3427 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3428 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3429 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3430 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3431 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3432 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3433 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3434 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3435 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3436 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3437 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3438 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3439 struct cp_control *cpc); 3440 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3441 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3442 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3443 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3444 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3445 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3446 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3447 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3448 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3449 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3450 unsigned int *newseg, bool new_sec, int dir); 3451 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3452 unsigned int start, unsigned int end); 3453 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3454 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3455 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3456 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3457 struct cp_control *cpc); 3458 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3459 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3460 block_t blk_addr); 3461 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3462 enum iostat_type io_type); 3463 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3464 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3465 struct f2fs_io_info *fio); 3466 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3467 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3468 block_t old_blkaddr, block_t new_blkaddr, 3469 bool recover_curseg, bool recover_newaddr, 3470 bool from_gc); 3471 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3472 block_t old_addr, block_t new_addr, 3473 unsigned char version, bool recover_curseg, 3474 bool recover_newaddr); 3475 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3476 block_t old_blkaddr, block_t *new_blkaddr, 3477 struct f2fs_summary *sum, int type, 3478 struct f2fs_io_info *fio); 3479 void f2fs_wait_on_page_writeback(struct page *page, 3480 enum page_type type, bool ordered, bool locked); 3481 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3482 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3483 block_t len); 3484 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3485 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3486 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3487 unsigned int val, int alloc); 3488 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3489 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3490 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3491 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3492 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3493 int __init f2fs_create_segment_manager_caches(void); 3494 void f2fs_destroy_segment_manager_caches(void); 3495 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3496 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3497 enum page_type type, enum temp_type temp); 3498 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3499 unsigned int segno); 3500 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3501 unsigned int segno); 3502 3503 /* 3504 * checkpoint.c 3505 */ 3506 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3507 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3508 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3509 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3510 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3511 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3512 block_t blkaddr, int type); 3513 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3514 int type, bool sync); 3515 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3516 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3517 long nr_to_write, enum iostat_type io_type); 3518 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3519 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3520 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3521 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3522 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3523 unsigned int devidx, int type); 3524 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3525 unsigned int devidx, int type); 3526 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3527 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3528 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3529 void f2fs_add_orphan_inode(struct inode *inode); 3530 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3531 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3532 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3533 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3534 void f2fs_remove_dirty_inode(struct inode *inode); 3535 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3536 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3537 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3538 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3539 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3540 int __init f2fs_create_checkpoint_caches(void); 3541 void f2fs_destroy_checkpoint_caches(void); 3542 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3543 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3544 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3545 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3546 3547 /* 3548 * data.c 3549 */ 3550 int __init f2fs_init_bioset(void); 3551 void f2fs_destroy_bioset(void); 3552 int f2fs_init_bio_entry_cache(void); 3553 void f2fs_destroy_bio_entry_cache(void); 3554 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3555 struct bio *bio, enum page_type type); 3556 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3557 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3558 struct inode *inode, struct page *page, 3559 nid_t ino, enum page_type type); 3560 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3561 struct bio **bio, struct page *page); 3562 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3563 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3564 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3565 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3566 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3567 block_t blk_addr, struct bio *bio); 3568 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3569 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3570 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3571 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3572 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3573 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3574 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3575 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3576 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3577 int op_flags, bool for_write); 3578 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3579 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3580 bool for_write); 3581 struct page *f2fs_get_new_data_page(struct inode *inode, 3582 struct page *ipage, pgoff_t index, bool new_i_size); 3583 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3584 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3585 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3586 int create, int flag); 3587 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3588 u64 start, u64 len); 3589 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3590 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3591 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3592 int f2fs_write_single_data_page(struct page *page, int *submitted, 3593 struct bio **bio, sector_t *last_block, 3594 struct writeback_control *wbc, 3595 enum iostat_type io_type, 3596 int compr_blocks, bool allow_balance); 3597 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3598 unsigned int length); 3599 int f2fs_release_page(struct page *page, gfp_t wait); 3600 #ifdef CONFIG_MIGRATION 3601 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3602 struct page *page, enum migrate_mode mode); 3603 #endif 3604 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3605 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3606 int f2fs_init_post_read_processing(void); 3607 void f2fs_destroy_post_read_processing(void); 3608 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3609 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3610 3611 /* 3612 * gc.c 3613 */ 3614 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3615 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3616 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3617 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force, 3618 unsigned int segno); 3619 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3620 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3621 int __init f2fs_create_garbage_collection_cache(void); 3622 void f2fs_destroy_garbage_collection_cache(void); 3623 3624 /* 3625 * recovery.c 3626 */ 3627 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3628 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3629 int __init f2fs_create_recovery_cache(void); 3630 void f2fs_destroy_recovery_cache(void); 3631 3632 /* 3633 * debug.c 3634 */ 3635 #ifdef CONFIG_F2FS_STAT_FS 3636 struct f2fs_stat_info { 3637 struct list_head stat_list; 3638 struct f2fs_sb_info *sbi; 3639 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3640 int main_area_segs, main_area_sections, main_area_zones; 3641 unsigned long long hit_largest, hit_cached, hit_rbtree; 3642 unsigned long long hit_total, total_ext; 3643 int ext_tree, zombie_tree, ext_node; 3644 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3645 int ndirty_data, ndirty_qdata; 3646 int inmem_pages; 3647 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3648 int nats, dirty_nats, sits, dirty_sits; 3649 int free_nids, avail_nids, alloc_nids; 3650 int total_count, utilization; 3651 int bg_gc, nr_wb_cp_data, nr_wb_data; 3652 int nr_rd_data, nr_rd_node, nr_rd_meta; 3653 int nr_dio_read, nr_dio_write; 3654 unsigned int io_skip_bggc, other_skip_bggc; 3655 int nr_flushing, nr_flushed, flush_list_empty; 3656 int nr_discarding, nr_discarded; 3657 int nr_discard_cmd; 3658 unsigned int undiscard_blks; 3659 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3660 unsigned int cur_ckpt_time, peak_ckpt_time; 3661 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3662 int compr_inode; 3663 unsigned long long compr_blocks; 3664 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3665 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3666 unsigned int bimodal, avg_vblocks; 3667 int util_free, util_valid, util_invalid; 3668 int rsvd_segs, overp_segs; 3669 int dirty_count, node_pages, meta_pages; 3670 int prefree_count, call_count, cp_count, bg_cp_count; 3671 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3672 int bg_node_segs, bg_data_segs; 3673 int tot_blks, data_blks, node_blks; 3674 int bg_data_blks, bg_node_blks; 3675 unsigned long long skipped_atomic_files[2]; 3676 int curseg[NR_CURSEG_TYPE]; 3677 int cursec[NR_CURSEG_TYPE]; 3678 int curzone[NR_CURSEG_TYPE]; 3679 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3680 unsigned int full_seg[NR_CURSEG_TYPE]; 3681 unsigned int valid_blks[NR_CURSEG_TYPE]; 3682 3683 unsigned int meta_count[META_MAX]; 3684 unsigned int segment_count[2]; 3685 unsigned int block_count[2]; 3686 unsigned int inplace_count; 3687 unsigned long long base_mem, cache_mem, page_mem; 3688 }; 3689 3690 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3691 { 3692 return (struct f2fs_stat_info *)sbi->stat_info; 3693 } 3694 3695 #define stat_inc_cp_count(si) ((si)->cp_count++) 3696 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3697 #define stat_inc_call_count(si) ((si)->call_count++) 3698 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3699 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3700 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3701 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3702 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3703 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3704 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3705 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3706 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3707 #define stat_inc_inline_xattr(inode) \ 3708 do { \ 3709 if (f2fs_has_inline_xattr(inode)) \ 3710 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3711 } while (0) 3712 #define stat_dec_inline_xattr(inode) \ 3713 do { \ 3714 if (f2fs_has_inline_xattr(inode)) \ 3715 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3716 } while (0) 3717 #define stat_inc_inline_inode(inode) \ 3718 do { \ 3719 if (f2fs_has_inline_data(inode)) \ 3720 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3721 } while (0) 3722 #define stat_dec_inline_inode(inode) \ 3723 do { \ 3724 if (f2fs_has_inline_data(inode)) \ 3725 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3726 } while (0) 3727 #define stat_inc_inline_dir(inode) \ 3728 do { \ 3729 if (f2fs_has_inline_dentry(inode)) \ 3730 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3731 } while (0) 3732 #define stat_dec_inline_dir(inode) \ 3733 do { \ 3734 if (f2fs_has_inline_dentry(inode)) \ 3735 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3736 } while (0) 3737 #define stat_inc_compr_inode(inode) \ 3738 do { \ 3739 if (f2fs_compressed_file(inode)) \ 3740 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3741 } while (0) 3742 #define stat_dec_compr_inode(inode) \ 3743 do { \ 3744 if (f2fs_compressed_file(inode)) \ 3745 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3746 } while (0) 3747 #define stat_add_compr_blocks(inode, blocks) \ 3748 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3749 #define stat_sub_compr_blocks(inode, blocks) \ 3750 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3751 #define stat_inc_meta_count(sbi, blkaddr) \ 3752 do { \ 3753 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3754 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3755 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3756 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3757 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3758 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3759 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3760 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3761 } while (0) 3762 #define stat_inc_seg_type(sbi, curseg) \ 3763 ((sbi)->segment_count[(curseg)->alloc_type]++) 3764 #define stat_inc_block_count(sbi, curseg) \ 3765 ((sbi)->block_count[(curseg)->alloc_type]++) 3766 #define stat_inc_inplace_blocks(sbi) \ 3767 (atomic_inc(&(sbi)->inplace_count)) 3768 #define stat_update_max_atomic_write(inode) \ 3769 do { \ 3770 int cur = F2FS_I_SB(inode)->atomic_files; \ 3771 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3772 if (cur > max) \ 3773 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3774 } while (0) 3775 #define stat_inc_volatile_write(inode) \ 3776 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3777 #define stat_dec_volatile_write(inode) \ 3778 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3779 #define stat_update_max_volatile_write(inode) \ 3780 do { \ 3781 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3782 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3783 if (cur > max) \ 3784 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3785 } while (0) 3786 #define stat_inc_seg_count(sbi, type, gc_type) \ 3787 do { \ 3788 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3789 si->tot_segs++; \ 3790 if ((type) == SUM_TYPE_DATA) { \ 3791 si->data_segs++; \ 3792 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3793 } else { \ 3794 si->node_segs++; \ 3795 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3796 } \ 3797 } while (0) 3798 3799 #define stat_inc_tot_blk_count(si, blks) \ 3800 ((si)->tot_blks += (blks)) 3801 3802 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3803 do { \ 3804 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3805 stat_inc_tot_blk_count(si, blks); \ 3806 si->data_blks += (blks); \ 3807 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3808 } while (0) 3809 3810 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3811 do { \ 3812 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3813 stat_inc_tot_blk_count(si, blks); \ 3814 si->node_blks += (blks); \ 3815 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3816 } while (0) 3817 3818 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3819 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3820 void __init f2fs_create_root_stats(void); 3821 void f2fs_destroy_root_stats(void); 3822 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3823 #else 3824 #define stat_inc_cp_count(si) do { } while (0) 3825 #define stat_inc_bg_cp_count(si) do { } while (0) 3826 #define stat_inc_call_count(si) do { } while (0) 3827 #define stat_inc_bggc_count(si) do { } while (0) 3828 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3829 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3830 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3831 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3832 #define stat_inc_total_hit(sbi) do { } while (0) 3833 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3834 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3835 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3836 #define stat_inc_inline_xattr(inode) do { } while (0) 3837 #define stat_dec_inline_xattr(inode) do { } while (0) 3838 #define stat_inc_inline_inode(inode) do { } while (0) 3839 #define stat_dec_inline_inode(inode) do { } while (0) 3840 #define stat_inc_inline_dir(inode) do { } while (0) 3841 #define stat_dec_inline_dir(inode) do { } while (0) 3842 #define stat_inc_compr_inode(inode) do { } while (0) 3843 #define stat_dec_compr_inode(inode) do { } while (0) 3844 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3845 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3846 #define stat_update_max_atomic_write(inode) do { } while (0) 3847 #define stat_inc_volatile_write(inode) do { } while (0) 3848 #define stat_dec_volatile_write(inode) do { } while (0) 3849 #define stat_update_max_volatile_write(inode) do { } while (0) 3850 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3851 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3852 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3853 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3854 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3855 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3856 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3857 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3858 3859 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3860 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3861 static inline void __init f2fs_create_root_stats(void) { } 3862 static inline void f2fs_destroy_root_stats(void) { } 3863 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 3864 #endif 3865 3866 extern const struct file_operations f2fs_dir_operations; 3867 extern const struct file_operations f2fs_file_operations; 3868 extern const struct inode_operations f2fs_file_inode_operations; 3869 extern const struct address_space_operations f2fs_dblock_aops; 3870 extern const struct address_space_operations f2fs_node_aops; 3871 extern const struct address_space_operations f2fs_meta_aops; 3872 extern const struct inode_operations f2fs_dir_inode_operations; 3873 extern const struct inode_operations f2fs_symlink_inode_operations; 3874 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3875 extern const struct inode_operations f2fs_special_inode_operations; 3876 extern struct kmem_cache *f2fs_inode_entry_slab; 3877 3878 /* 3879 * inline.c 3880 */ 3881 bool f2fs_may_inline_data(struct inode *inode); 3882 bool f2fs_may_inline_dentry(struct inode *inode); 3883 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3884 void f2fs_truncate_inline_inode(struct inode *inode, 3885 struct page *ipage, u64 from); 3886 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3887 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3888 int f2fs_convert_inline_inode(struct inode *inode); 3889 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3890 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3891 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3892 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3893 const struct f2fs_filename *fname, 3894 struct page **res_page); 3895 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3896 struct page *ipage); 3897 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 3898 struct inode *inode, nid_t ino, umode_t mode); 3899 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3900 struct page *page, struct inode *dir, 3901 struct inode *inode); 3902 bool f2fs_empty_inline_dir(struct inode *dir); 3903 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3904 struct fscrypt_str *fstr); 3905 int f2fs_inline_data_fiemap(struct inode *inode, 3906 struct fiemap_extent_info *fieinfo, 3907 __u64 start, __u64 len); 3908 3909 /* 3910 * shrinker.c 3911 */ 3912 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3913 struct shrink_control *sc); 3914 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3915 struct shrink_control *sc); 3916 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3917 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3918 3919 /* 3920 * extent_cache.c 3921 */ 3922 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3923 struct rb_entry *cached_re, unsigned int ofs); 3924 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 3925 struct rb_root_cached *root, 3926 struct rb_node **parent, 3927 unsigned long long key, bool *left_most); 3928 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3929 struct rb_root_cached *root, 3930 struct rb_node **parent, 3931 unsigned int ofs, bool *leftmost); 3932 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3933 struct rb_entry *cached_re, unsigned int ofs, 3934 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3935 struct rb_node ***insert_p, struct rb_node **insert_parent, 3936 bool force, bool *leftmost); 3937 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3938 struct rb_root_cached *root, bool check_key); 3939 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3940 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 3941 void f2fs_drop_extent_tree(struct inode *inode); 3942 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3943 void f2fs_destroy_extent_tree(struct inode *inode); 3944 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3945 struct extent_info *ei); 3946 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3947 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3948 pgoff_t fofs, block_t blkaddr, unsigned int len); 3949 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3950 int __init f2fs_create_extent_cache(void); 3951 void f2fs_destroy_extent_cache(void); 3952 3953 /* 3954 * sysfs.c 3955 */ 3956 int __init f2fs_init_sysfs(void); 3957 void f2fs_exit_sysfs(void); 3958 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3959 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3960 3961 /* verity.c */ 3962 extern const struct fsverity_operations f2fs_verityops; 3963 3964 /* 3965 * crypto support 3966 */ 3967 static inline bool f2fs_encrypted_file(struct inode *inode) 3968 { 3969 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 3970 } 3971 3972 static inline void f2fs_set_encrypted_inode(struct inode *inode) 3973 { 3974 #ifdef CONFIG_FS_ENCRYPTION 3975 file_set_encrypt(inode); 3976 f2fs_set_inode_flags(inode); 3977 #endif 3978 } 3979 3980 /* 3981 * Returns true if the reads of the inode's data need to undergo some 3982 * postprocessing step, like decryption or authenticity verification. 3983 */ 3984 static inline bool f2fs_post_read_required(struct inode *inode) 3985 { 3986 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 3987 f2fs_compressed_file(inode); 3988 } 3989 3990 /* 3991 * compress.c 3992 */ 3993 #ifdef CONFIG_F2FS_FS_COMPRESSION 3994 bool f2fs_is_compressed_page(struct page *page); 3995 struct page *f2fs_compress_control_page(struct page *page); 3996 int f2fs_prepare_compress_overwrite(struct inode *inode, 3997 struct page **pagep, pgoff_t index, void **fsdata); 3998 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 3999 pgoff_t index, unsigned copied); 4000 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4001 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4002 bool f2fs_is_compress_backend_ready(struct inode *inode); 4003 int f2fs_init_compress_mempool(void); 4004 void f2fs_destroy_compress_mempool(void); 4005 void f2fs_end_read_compressed_page(struct page *page, bool failed); 4006 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4007 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4008 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4009 int f2fs_write_multi_pages(struct compress_ctx *cc, 4010 int *submitted, 4011 struct writeback_control *wbc, 4012 enum iostat_type io_type); 4013 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4014 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4015 unsigned nr_pages, sector_t *last_block_in_bio, 4016 bool is_readahead, bool for_write); 4017 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4018 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed); 4019 void f2fs_put_page_dic(struct page *page); 4020 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4021 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4022 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4023 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4024 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4025 int __init f2fs_init_compress_cache(void); 4026 void f2fs_destroy_compress_cache(void); 4027 #define inc_compr_inode_stat(inode) \ 4028 do { \ 4029 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4030 sbi->compr_new_inode++; \ 4031 } while (0) 4032 #define add_compr_block_stat(inode, blocks) \ 4033 do { \ 4034 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4035 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4036 sbi->compr_written_block += blocks; \ 4037 sbi->compr_saved_block += diff; \ 4038 } while (0) 4039 #else 4040 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4041 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4042 { 4043 if (!f2fs_compressed_file(inode)) 4044 return true; 4045 /* not support compression */ 4046 return false; 4047 } 4048 static inline struct page *f2fs_compress_control_page(struct page *page) 4049 { 4050 WARN_ON_ONCE(1); 4051 return ERR_PTR(-EINVAL); 4052 } 4053 static inline int f2fs_init_compress_mempool(void) { return 0; } 4054 static inline void f2fs_destroy_compress_mempool(void) { } 4055 static inline void f2fs_end_read_compressed_page(struct page *page, bool failed) 4056 { 4057 WARN_ON_ONCE(1); 4058 } 4059 static inline void f2fs_put_page_dic(struct page *page) 4060 { 4061 WARN_ON_ONCE(1); 4062 } 4063 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4064 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4065 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4066 static inline void f2fs_destroy_compress_cache(void) { } 4067 #define inc_compr_inode_stat(inode) do { } while (0) 4068 #endif 4069 4070 static inline void set_compress_context(struct inode *inode) 4071 { 4072 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4073 4074 F2FS_I(inode)->i_compress_algorithm = 4075 F2FS_OPTION(sbi).compress_algorithm; 4076 F2FS_I(inode)->i_log_cluster_size = 4077 F2FS_OPTION(sbi).compress_log_size; 4078 F2FS_I(inode)->i_compress_flag = 4079 F2FS_OPTION(sbi).compress_chksum ? 4080 1 << COMPRESS_CHKSUM : 0; 4081 F2FS_I(inode)->i_cluster_size = 4082 1 << F2FS_I(inode)->i_log_cluster_size; 4083 if (F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 && 4084 F2FS_OPTION(sbi).compress_level) 4085 F2FS_I(inode)->i_compress_flag |= 4086 F2FS_OPTION(sbi).compress_level << 4087 COMPRESS_LEVEL_OFFSET; 4088 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4089 set_inode_flag(inode, FI_COMPRESSED_FILE); 4090 stat_inc_compr_inode(inode); 4091 inc_compr_inode_stat(inode); 4092 f2fs_mark_inode_dirty_sync(inode, true); 4093 } 4094 4095 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4096 { 4097 struct f2fs_inode_info *fi = F2FS_I(inode); 4098 4099 if (!f2fs_compressed_file(inode)) 4100 return true; 4101 if (S_ISREG(inode->i_mode) && 4102 (get_dirty_pages(inode) || atomic_read(&fi->i_compr_blocks))) 4103 return false; 4104 4105 fi->i_flags &= ~F2FS_COMPR_FL; 4106 stat_dec_compr_inode(inode); 4107 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4108 f2fs_mark_inode_dirty_sync(inode, true); 4109 return true; 4110 } 4111 4112 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4113 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4114 { \ 4115 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4116 } 4117 4118 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4119 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4120 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4121 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4122 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4123 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4124 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4125 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4126 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4127 F2FS_FEATURE_FUNCS(verity, VERITY); 4128 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4129 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4130 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4131 4132 #ifdef CONFIG_BLK_DEV_ZONED 4133 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4134 block_t blkaddr) 4135 { 4136 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4137 4138 return test_bit(zno, FDEV(devi).blkz_seq); 4139 } 4140 #endif 4141 4142 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4143 { 4144 return f2fs_sb_has_blkzoned(sbi); 4145 } 4146 4147 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4148 { 4149 return blk_queue_discard(bdev_get_queue(bdev)) || 4150 bdev_is_zoned(bdev); 4151 } 4152 4153 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4154 { 4155 int i; 4156 4157 if (!f2fs_is_multi_device(sbi)) 4158 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4159 4160 for (i = 0; i < sbi->s_ndevs; i++) 4161 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4162 return true; 4163 return false; 4164 } 4165 4166 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4167 { 4168 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4169 f2fs_hw_should_discard(sbi); 4170 } 4171 4172 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4173 { 4174 int i; 4175 4176 if (!f2fs_is_multi_device(sbi)) 4177 return bdev_read_only(sbi->sb->s_bdev); 4178 4179 for (i = 0; i < sbi->s_ndevs; i++) 4180 if (bdev_read_only(FDEV(i).bdev)) 4181 return true; 4182 return false; 4183 } 4184 4185 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4186 { 4187 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4188 } 4189 4190 static inline bool f2fs_may_compress(struct inode *inode) 4191 { 4192 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4193 f2fs_is_atomic_file(inode) || 4194 f2fs_is_volatile_file(inode)) 4195 return false; 4196 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4197 } 4198 4199 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4200 u64 blocks, bool add) 4201 { 4202 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4203 struct f2fs_inode_info *fi = F2FS_I(inode); 4204 4205 /* don't update i_compr_blocks if saved blocks were released */ 4206 if (!add && !atomic_read(&fi->i_compr_blocks)) 4207 return; 4208 4209 if (add) { 4210 atomic_add(diff, &fi->i_compr_blocks); 4211 stat_add_compr_blocks(inode, diff); 4212 } else { 4213 atomic_sub(diff, &fi->i_compr_blocks); 4214 stat_sub_compr_blocks(inode, diff); 4215 } 4216 f2fs_mark_inode_dirty_sync(inode, true); 4217 } 4218 4219 static inline int block_unaligned_IO(struct inode *inode, 4220 struct kiocb *iocb, struct iov_iter *iter) 4221 { 4222 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4223 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4224 loff_t offset = iocb->ki_pos; 4225 unsigned long align = offset | iov_iter_alignment(iter); 4226 4227 return align & blocksize_mask; 4228 } 4229 4230 static inline int allow_outplace_dio(struct inode *inode, 4231 struct kiocb *iocb, struct iov_iter *iter) 4232 { 4233 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4234 int rw = iov_iter_rw(iter); 4235 4236 return (f2fs_lfs_mode(sbi) && (rw == WRITE) && 4237 !block_unaligned_IO(inode, iocb, iter)); 4238 } 4239 4240 static inline bool f2fs_force_buffered_io(struct inode *inode, 4241 struct kiocb *iocb, struct iov_iter *iter) 4242 { 4243 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4244 int rw = iov_iter_rw(iter); 4245 4246 if (f2fs_post_read_required(inode)) 4247 return true; 4248 if (f2fs_is_multi_device(sbi)) 4249 return true; 4250 /* 4251 * for blkzoned device, fallback direct IO to buffered IO, so 4252 * all IOs can be serialized by log-structured write. 4253 */ 4254 if (f2fs_sb_has_blkzoned(sbi)) 4255 return true; 4256 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4257 if (block_unaligned_IO(inode, iocb, iter)) 4258 return true; 4259 if (F2FS_IO_ALIGNED(sbi)) 4260 return true; 4261 } 4262 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 4263 return true; 4264 4265 return false; 4266 } 4267 4268 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4269 { 4270 return fsverity_active(inode) && 4271 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4272 } 4273 4274 #ifdef CONFIG_F2FS_FAULT_INJECTION 4275 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4276 unsigned int type); 4277 #else 4278 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4279 #endif 4280 4281 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4282 { 4283 #ifdef CONFIG_QUOTA 4284 if (f2fs_sb_has_quota_ino(sbi)) 4285 return true; 4286 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4287 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4288 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4289 return true; 4290 #endif 4291 return false; 4292 } 4293 4294 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4295 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4296 4297 #endif /* _LINUX_F2FS_H */ 4298